bibtex_url
stringlengths 41
53
| acl_proceedings
stringlengths 38
50
| bibtext
stringlengths 528
3.02k
| abstract
stringlengths 17
2.35k
| authors
sequencelengths 1
44
| title
stringlengths 18
190
| id
stringlengths 7
19
| arxiv_id
stringlengths 10
10
⌀ | GitHub
sequencelengths 1
1
| paper_page
stringclasses 528
values | n_linked_authors
int64 -1
15
| upvotes
int64 -1
77
| num_comments
int64 -1
10
| n_authors
int64 -1
52
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
15
| Spaces
sequencelengths 0
46
| paper_page_exists_pre_conf
int64 0
1
| type
stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://aclanthology.org/2023.findings-emnlp.19.bib | https://aclanthology.org/2023.findings-emnlp.19/ | @inproceedings{mccarthy-etal-2023-long,
title = "Long-Form Speech Translation through Segmentation with Finite-State Decoding Constraints on Large Language Models",
author = "McCarthy, Arya and
Zhang, Hao and
Kumar, Shankar and
Stahlberg, Felix and
Wu, Ke",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.19",
doi = "10.18653/v1/2023.findings-emnlp.19",
pages = "247--257",
abstract = "One challenge in speech translation is that plenty of spoken content is long-form, but short units are necessary for obtaining high-quality translations. To address this mismatch, we adapt large language models (LLMs) to split long ASR transcripts into segments that can be independently translated so as to maximize the overall translation quality. We overcome the tendency of hallucination in LLMs by incorporating finite-state constraints during decoding; these eliminate invalid outputs without requiring additional training. We discover that LLMs are adaptable to transcripts containing ASR errors through prompt-tuning or fine-tuning. Relative to a state-of-the-art automatic punctuation baseline, our best LLM improves the average BLEU by 2.9 points for English{--}German, English{--}Spanish, and English{--}Arabic TED talk translation in 9 test sets, just by improving segmentation.",
}
| One challenge in speech translation is that plenty of spoken content is long-form, but short units are necessary for obtaining high-quality translations. To address this mismatch, we adapt large language models (LLMs) to split long ASR transcripts into segments that can be independently translated so as to maximize the overall translation quality. We overcome the tendency of hallucination in LLMs by incorporating finite-state constraints during decoding; these eliminate invalid outputs without requiring additional training. We discover that LLMs are adaptable to transcripts containing ASR errors through prompt-tuning or fine-tuning. Relative to a state-of-the-art automatic punctuation baseline, our best LLM improves the average BLEU by 2.9 points for English{--}German, English{--}Spanish, and English{--}Arabic TED talk translation in 9 test sets, just by improving segmentation. | [
"McCarthy, Arya",
"Zhang, Hao",
"Kumar, Shankar",
"Stahlberg, Felix",
"Wu, Ke"
] | Long-Form Speech Translation through Segmentation with Finite-State Decoding Constraints on Large Language Models | findings-emnlp.19 | 2310.13678 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.20.bib | https://aclanthology.org/2023.findings-emnlp.20/ | @inproceedings{wang-etal-2023-temp,
title = "Re-Temp: Relation-Aware Temporal Representation Learning for Temporal Knowledge Graph Completion",
author = "Wang, Kunze and
Han, Caren and
Poon, Josiah",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.20",
doi = "10.18653/v1/2023.findings-emnlp.20",
pages = "258--269",
abstract = "Temporal Knowledge Graph Completion (TKGC) under the extrapolation setting aims to predict the missing entity from a fact in the future, posing a challenge that aligns more closely with real-world prediction problems. Existing research mostly encodes entities and relations using sequential graph neural networks applied to recent snapshots. However, these approaches tend to overlook the ability to skip irrelevant snapshots according to entity-related relations in the query and disregard the importance of explicit temporal information. To address this, we propose our model, Re-Temp (Relation-Aware Temporal Representation Learning), which leverages explicit temporal embedding as input and incorporates skip information flow after each timestamp to skip unnecessary information for prediction. Additionally, we introduce a two-phase forward propagation method to prevent information leakage. Through the evaluation on six TKGC (extrapolation) datasets, we demonstrate that our model outperforms all eight recent state-of-the-art models by a significant margin.",
}
| Temporal Knowledge Graph Completion (TKGC) under the extrapolation setting aims to predict the missing entity from a fact in the future, posing a challenge that aligns more closely with real-world prediction problems. Existing research mostly encodes entities and relations using sequential graph neural networks applied to recent snapshots. However, these approaches tend to overlook the ability to skip irrelevant snapshots according to entity-related relations in the query and disregard the importance of explicit temporal information. To address this, we propose our model, Re-Temp (Relation-Aware Temporal Representation Learning), which leverages explicit temporal embedding as input and incorporates skip information flow after each timestamp to skip unnecessary information for prediction. Additionally, we introduce a two-phase forward propagation method to prevent information leakage. Through the evaluation on six TKGC (extrapolation) datasets, we demonstrate that our model outperforms all eight recent state-of-the-art models by a significant margin. | [
"Wang, Kunze",
"Han, Caren",
"Poon, Josiah"
] | Re-Temp: Relation-Aware Temporal Representation Learning for Temporal Knowledge Graph Completion | findings-emnlp.20 | 2310.15722 | [
"https://github.com/adlnlp/re-temp"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.21.bib | https://aclanthology.org/2023.findings-emnlp.21/ | @inproceedings{ye-etal-2023-rethinkingtmsc,
title = "{R}ethinking{TMSC}: An Empirical Study for Target-Oriented Multimodal Sentiment Classification",
author = "Ye, Junjie and
Zhou, Jie and
Tian, Junfeng and
Wang, Rui and
Zhang, Qi and
Gui, Tao and
Huang, Xuanjing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.21",
doi = "10.18653/v1/2023.findings-emnlp.21",
pages = "270--277",
abstract = "Recently, Target-oriented Multimodal Sentiment Classification (TMSC) has gained significant attention among scholars. However, current multimodal models have reached a performance bottleneck. To investigate the causes of this problem, we perform extensive empirical evaluation and in-depth analysis of the datasets to answer the following questions: **Q1**: Are the modalities equally important for TMSC? **Q2**: Which multimodal fusion modules are more effective? **Q3**: Do existing datasets adequately support the research? Our experiments and analyses reveal that the current TMSC systems primarily rely on the textual modality, as most of targets{'} sentiments can be determined *solely* by text. Consequently, we point out several directions to work on for the TMSC task in terms of model design and dataset construction. The code and data can be found in https://github.com/Junjie-Ye/RethinkingTMSC.",
}
| Recently, Target-oriented Multimodal Sentiment Classification (TMSC) has gained significant attention among scholars. However, current multimodal models have reached a performance bottleneck. To investigate the causes of this problem, we perform extensive empirical evaluation and in-depth analysis of the datasets to answer the following questions: **Q1**: Are the modalities equally important for TMSC? **Q2**: Which multimodal fusion modules are more effective? **Q3**: Do existing datasets adequately support the research? Our experiments and analyses reveal that the current TMSC systems primarily rely on the textual modality, as most of targets{'} sentiments can be determined *solely* by text. Consequently, we point out several directions to work on for the TMSC task in terms of model design and dataset construction. The code and data can be found in https://github.com/Junjie-Ye/RethinkingTMSC. | [
"Ye, Junjie",
"Zhou, Jie",
"Tian, Junfeng",
"Wang, Rui",
"Zhang, Qi",
"Gui, Tao",
"Huang, Xuanjing"
] | RethinkingTMSC: An Empirical Study for Target-Oriented Multimodal Sentiment Classification | findings-emnlp.21 | 2310.09596 | [
"https://github.com/junjie-ye/rethinkingtmsc"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.22.bib | https://aclanthology.org/2023.findings-emnlp.22/ | @inproceedings{shi-etal-2023-lexical,
title = "Lexical Entrainment for Conversational Systems",
author = "Shi, Zhengxiang and
Sen, Procheta and
Lipani, Aldo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.22",
doi = "10.18653/v1/2023.findings-emnlp.22",
pages = "278--293",
abstract = "Conversational agents have become ubiquitous in assisting with daily tasks, and are expected to possess human-like features. One such feature is lexical entrainment (LE), a phenomenon in which speakers in human-human conversations tend to naturally and subconsciously align their lexical choices with those of their interlocutors, leading to more successful and engaging conversations. As an example, if a digital assistant replies {``}Your appointment for Jinling Noodle Pub is at 7 pm{''} to the question {``}When is my reservation for Jinling Noodle Bar today?{''}, it may feel as though the assistant is trying to correct the speaker, whereas a response of {``}Your reservation for Jinling Noodle Baris at 7 pm{''} would likely be perceived as more positive. This highlights the importance of LE in establishing a shared terminology for maximum clarity and reducing ambiguity in conversations. However, we demonstrate in this work that current response generation models do not adequately address this crucial human-like phenomenon. To address this, we propose a new dataset, named MultiWOZ-ENTR, and a measure for LE for conversational systems. Additionally, we suggest a way to explicitly integrate LE into conversational systems with two new tasks, a LE extraction task and a LE generation task. We also present two baseline approaches for the LE extraction task, which aim to detect LE expressions from dialogue contexts",
}
| Conversational agents have become ubiquitous in assisting with daily tasks, and are expected to possess human-like features. One such feature is lexical entrainment (LE), a phenomenon in which speakers in human-human conversations tend to naturally and subconsciously align their lexical choices with those of their interlocutors, leading to more successful and engaging conversations. As an example, if a digital assistant replies {``}Your appointment for Jinling Noodle Pub is at 7 pm{''} to the question {``}When is my reservation for Jinling Noodle Bar today?{''}, it may feel as though the assistant is trying to correct the speaker, whereas a response of {``}Your reservation for Jinling Noodle Baris at 7 pm{''} would likely be perceived as more positive. This highlights the importance of LE in establishing a shared terminology for maximum clarity and reducing ambiguity in conversations. However, we demonstrate in this work that current response generation models do not adequately address this crucial human-like phenomenon. To address this, we propose a new dataset, named MultiWOZ-ENTR, and a measure for LE for conversational systems. Additionally, we suggest a way to explicitly integrate LE into conversational systems with two new tasks, a LE extraction task and a LE generation task. We also present two baseline approaches for the LE extraction task, which aim to detect LE expressions from dialogue contexts | [
"Shi, Zhengxiang",
"Sen, Procheta",
"Lipani, Aldo"
] | Lexical Entrainment for Conversational Systems | findings-emnlp.22 | 2310.09651 | [
"https://github.com/zhengxiangshi/lexicalentrainment"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.23.bib | https://aclanthology.org/2023.findings-emnlp.23/ | @inproceedings{shi-etal-2023-autoreply,
title = "{A}uto{R}eply: Detecting Nonsense in Dialogue with Discriminative Replies",
author = "Shi, Weiyan and
Dinan, Emily and
Renduchintala, Adi and
Fried, Daniel and
Jacob, Athul and
Yu, Zhou and
Lewis, Mike",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.23",
doi = "10.18653/v1/2023.findings-emnlp.23",
pages = "294--309",
abstract = "We show that dialogue models can detect errors in their own messages, by calculating the likelihood of replies that are indicative of poor messages. For example, if an agent believes its partner is likely to respond {``}I don{'}t understand{''} to a candidate message, that message may not make sense, so an alternative message should be chosen. We evaluate our approach on a dataset from the game Diplomacy, which contains long dialogues richly grounded in the game state, on which existing models make many errors. We first show that hand-crafted replies can be effective for the task of detecting nonsense in applications as complex as Diplomacy. We then design AutoReply, an algorithm to search for such discriminative replies automatically, given a small number of annotated dialogue examples. We find that AutoReply-generated replies outperform handcrafted replies and perform on par with supervised learning approaches.",
}
| We show that dialogue models can detect errors in their own messages, by calculating the likelihood of replies that are indicative of poor messages. For example, if an agent believes its partner is likely to respond {``}I don{'}t understand{''} to a candidate message, that message may not make sense, so an alternative message should be chosen. We evaluate our approach on a dataset from the game Diplomacy, which contains long dialogues richly grounded in the game state, on which existing models make many errors. We first show that hand-crafted replies can be effective for the task of detecting nonsense in applications as complex as Diplomacy. We then design AutoReply, an algorithm to search for such discriminative replies automatically, given a small number of annotated dialogue examples. We find that AutoReply-generated replies outperform handcrafted replies and perform on par with supervised learning approaches. | [
"Shi, Weiyan",
"Dinan, Emily",
"Renduchintala, Adi",
"Fried, Daniel",
"Jacob, Athul",
"Yu, Zhou",
"Lewis, Mike"
] | AutoReply: Detecting Nonsense in Dialogue with Discriminative Replies | findings-emnlp.23 | 2211.12615 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.24.bib | https://aclanthology.org/2023.findings-emnlp.24/ | @inproceedings{fetahu-etal-2023-follow,
title = "Follow-on Question Suggestion via Voice Hints for Voice Assistants",
author = "Fetahu, Besnik and
Faustini, Pedro and
Fang, Anjie and
Castellucci, Giuseppe and
Rokhlenko, Oleg and
Malmasi, Shervin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.24",
doi = "10.18653/v1/2023.findings-emnlp.24",
pages = "310--325",
abstract = "The adoption of voice assistants like Alexa or Siri has grown rapidly, allowing users to instantly access information via voice search. Query suggestion is a standard feature of screen-based search experiences, allowing users to explore additional topics. However, this is not trivial to implement in voice-based settings. To enable this, we tackle the novel task of suggesting questions with compact and natural voice hints to allow users to ask follow-up questions. We define the task, ground it in syntactic theory and outline linguistic desiderata for spoken hints. We propose baselines and an approach using sequence-to-sequence Transformers to generate spoken hints from a list of questions. Using a new dataset of 6681 input questions and human written hints, we evaluated the models with automatic metrics and human evaluation. Results show that a naive approach of concatenating suggested questions creates poor voice hints. Our approach, which applies a linguistically-motivated pretraining task was strongly preferred by humans for producing the most natural hints.",
}
| The adoption of voice assistants like Alexa or Siri has grown rapidly, allowing users to instantly access information via voice search. Query suggestion is a standard feature of screen-based search experiences, allowing users to explore additional topics. However, this is not trivial to implement in voice-based settings. To enable this, we tackle the novel task of suggesting questions with compact and natural voice hints to allow users to ask follow-up questions. We define the task, ground it in syntactic theory and outline linguistic desiderata for spoken hints. We propose baselines and an approach using sequence-to-sequence Transformers to generate spoken hints from a list of questions. Using a new dataset of 6681 input questions and human written hints, we evaluated the models with automatic metrics and human evaluation. Results show that a naive approach of concatenating suggested questions creates poor voice hints. Our approach, which applies a linguistically-motivated pretraining task was strongly preferred by humans for producing the most natural hints. | [
"Fetahu, Besnik",
"Faustini, Pedro",
"Fang, Anjie",
"Castellucci, Giuseppe",
"Rokhlenko, Oleg",
"Malmasi, Shervin"
] | Follow-on Question Suggestion via Voice Hints for Voice Assistants | findings-emnlp.24 | 2310.17034 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.25.bib | https://aclanthology.org/2023.findings-emnlp.25/ | @inproceedings{kim-etal-2023-bidirectional,
title = "Bidirectional Masked Self-attention and N-gram Span Attention for Constituency Parsing",
author = "Kim, Soohyeong and
Cho, Whanhee and
Kim, Minji and
Choi, Yong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.25",
doi = "10.18653/v1/2023.findings-emnlp.25",
pages = "326--338",
abstract = "Attention mechanisms have become a crucial aspect of deep learning, particularly in natural language processing (NLP) tasks. However, in tasks such as constituency parsing, attention mechanisms can lack the directional information needed to form sentence spans. To address this issue, we propose a Bidirectional masked and N-gram span Attention (BNA) model, which is designed by modifying the attention mechanisms to capture the explicit dependencies between each word and enhance the representation of the output span vectors. The proposed model achieves state-of-the-art performance on the Penn Treebank and Chinese Penn Treebank datasets, with F1 scores of 96.47 and 94.15, respectively. Ablation studies and analysis show that our proposed BNA model effectively captures sentence structure by contextualizing each word in a sentence through bidirectional dependencies and enhancing span representation.",
}
| Attention mechanisms have become a crucial aspect of deep learning, particularly in natural language processing (NLP) tasks. However, in tasks such as constituency parsing, attention mechanisms can lack the directional information needed to form sentence spans. To address this issue, we propose a Bidirectional masked and N-gram span Attention (BNA) model, which is designed by modifying the attention mechanisms to capture the explicit dependencies between each word and enhance the representation of the output span vectors. The proposed model achieves state-of-the-art performance on the Penn Treebank and Chinese Penn Treebank datasets, with F1 scores of 96.47 and 94.15, respectively. Ablation studies and analysis show that our proposed BNA model effectively captures sentence structure by contextualizing each word in a sentence through bidirectional dependencies and enhancing span representation. | [
"Kim, Soohyeong",
"Cho, Whanhee",
"Kim, Minji",
"Choi, Yong"
] | Bidirectional Masked Self-attention and N-gram Span Attention for Constituency Parsing | findings-emnlp.25 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.26.bib | https://aclanthology.org/2023.findings-emnlp.26/ | @inproceedings{chun-etal-2023-cr,
title = "{CR}-{COPEC}: Causal Rationale of Corporate Performance Changes to learn from Financial Reports",
author = "Chun, Ye and
Kwon, Sunjae and
Sohn, Kyunghwan and
Sung, Nakwon and
Lee, Junyoup and
Seo, Byoung and
Compher, Kevin and
Hwang, Seung-won and
Choi, Jaesik",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.26",
doi = "10.18653/v1/2023.findings-emnlp.26",
pages = "339--355",
abstract = "In this paper, we introduce CR-COPEC called Causal Rationale of Corporate Performance Changes from financial reports. This is a comprehensive large-scale domain-adaptation causal sentence dataset to detect financial performance changes of corporate. CR-COPEC contributes to two major achievements. First, it detects causal rationale from 10-K annual reports of the U.S. companies, which contain experts{'} causal analysis following accounting standards in a formal manner. This dataset can be widely used by both individual investors and analysts as material information resources for investing and decision-making without tremendous effort to read through all the documents. Second, it carefully considers different characteristics which affect the financial performance of companies in twelve industries. As a result, CR-COPEC can distinguish causal sentences in various industries by taking unique narratives in each industry into consideration. We also provide an extensive analysis of how well CR-COPEC dataset is constructed and suited for classifying target sentences as causal ones with respect to industry characteristics.",
}
| In this paper, we introduce CR-COPEC called Causal Rationale of Corporate Performance Changes from financial reports. This is a comprehensive large-scale domain-adaptation causal sentence dataset to detect financial performance changes of corporate. CR-COPEC contributes to two major achievements. First, it detects causal rationale from 10-K annual reports of the U.S. companies, which contain experts{'} causal analysis following accounting standards in a formal manner. This dataset can be widely used by both individual investors and analysts as material information resources for investing and decision-making without tremendous effort to read through all the documents. Second, it carefully considers different characteristics which affect the financial performance of companies in twelve industries. As a result, CR-COPEC can distinguish causal sentences in various industries by taking unique narratives in each industry into consideration. We also provide an extensive analysis of how well CR-COPEC dataset is constructed and suited for classifying target sentences as causal ones with respect to industry characteristics. | [
"Chun, Ye",
"Kwon, Sunjae",
"Sohn, Kyunghwan",
"Sung, Nakwon",
"Lee, Junyoup",
"Seo, Byoung",
"Compher, Kevin",
"Hwang, Seung-won",
"Choi, Jaesik"
] | CR-COPEC: Causal Rationale of Corporate Performance Changes to learn from Financial Reports | findings-emnlp.26 | 2310.16095 | [
"https://github.com/cr-copec/cr-copec"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.27.bib | https://aclanthology.org/2023.findings-emnlp.27/ | @inproceedings{ryu-2023-plausibility,
title = "Plausibility Processing in Transformer Language Models: Focusing on the Role of Attention Heads in {GPT}",
author = "Ryu, Soo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.27",
doi = "10.18653/v1/2023.findings-emnlp.27",
pages = "356--369",
abstract = "The goal of this paper is to explore how Transformer language models process semantic knowledge, especially regarding the plausibility of noun-verb relations. First, I demonstrate GPT2 exhibits a higher degree of similarity with humans in plausibility processing compared to other Transformer language models. Next, I delve into how knowledge of plausibility is contained within attention heads of GPT2 and how these heads causally contribute to GPT2{'}s plausibility processing ability. Through several experiments, it was found that: i) GPT2 has a number of attention heads that detect plausible noun-verb relationships; ii) these heads collectively contribute to the Transformer{'}s ability to process plausibility, albeit to varying degrees; and iii) attention heads{'} individual performance in detecting plausibility does not necessarily correlate with how much they contribute to GPT2{'}s plausibility processing ability.",
}
| The goal of this paper is to explore how Transformer language models process semantic knowledge, especially regarding the plausibility of noun-verb relations. First, I demonstrate GPT2 exhibits a higher degree of similarity with humans in plausibility processing compared to other Transformer language models. Next, I delve into how knowledge of plausibility is contained within attention heads of GPT2 and how these heads causally contribute to GPT2{'}s plausibility processing ability. Through several experiments, it was found that: i) GPT2 has a number of attention heads that detect plausible noun-verb relationships; ii) these heads collectively contribute to the Transformer{'}s ability to process plausibility, albeit to varying degrees; and iii) attention heads{'} individual performance in detecting plausibility does not necessarily correlate with how much they contribute to GPT2{'}s plausibility processing ability. | [
"Ryu, Soo"
] | Plausibility Processing in Transformer Language Models: Focusing on the Role of Attention Heads in GPT | findings-emnlp.27 | 2310.13824 | [
"https://github.com/soohyunryu/plausibility-processing-transformers"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.28.bib | https://aclanthology.org/2023.findings-emnlp.28/ | @inproceedings{gorinski-etal-2023-automatic,
title = "Automatic Unit Test Data Generation and Actor-Critic Reinforcement Learning for Code Synthesis",
author = "Gorinski, Philip and
Zimmer, Matthieu and
Lampouras, Gerasimos and
Deik, Derrick Goh Xin and
Iacobacci, Ignacio",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.28",
doi = "10.18653/v1/2023.findings-emnlp.28",
pages = "370--384",
abstract = "The advent of large pre-trained language models in the domain of Code Synthesis has shown remarkable performance on various benchmarks, treating the problem of Code Generation in a fashion similar to Natural Language Generation, trained with a Language Modelling (LM) objective. In addition, the property of programming language code being precisely evaluable with respect to its semantics {--} through the use of Unit Tests to check its functional correctness {--} lends itself to using Reinforcement Learning (RL) as a further training paradigm. Previous work has shown that RL can be applied as such to improve models{'} coding capabilities; however, such RL-based methods rely on a reward signal based on defined Unit Tests, which are much harder to obtain compared to the huge crawled code datasets used in LM objectives. In this work, we present a novel approach to automatically obtain data consisting of function signatures and associated Unit Tests, suitable for RL training of Code Synthesis models. We also introduce a straightforward, simple yet effective Actor-Critic RL training scheme and show that it, in conjunction with automatically generated training data, leads to improvement of a pre-trained code language model{'}s performance by up to 9.9{\%} improvement over the original underlying code synthesis LM, and up to 4.3{\%} over RL-based models trained with standard PPO or CodeRL.",
}
| The advent of large pre-trained language models in the domain of Code Synthesis has shown remarkable performance on various benchmarks, treating the problem of Code Generation in a fashion similar to Natural Language Generation, trained with a Language Modelling (LM) objective. In addition, the property of programming language code being precisely evaluable with respect to its semantics {--} through the use of Unit Tests to check its functional correctness {--} lends itself to using Reinforcement Learning (RL) as a further training paradigm. Previous work has shown that RL can be applied as such to improve models{'} coding capabilities; however, such RL-based methods rely on a reward signal based on defined Unit Tests, which are much harder to obtain compared to the huge crawled code datasets used in LM objectives. In this work, we present a novel approach to automatically obtain data consisting of function signatures and associated Unit Tests, suitable for RL training of Code Synthesis models. We also introduce a straightforward, simple yet effective Actor-Critic RL training scheme and show that it, in conjunction with automatically generated training data, leads to improvement of a pre-trained code language model{'}s performance by up to 9.9{\%} improvement over the original underlying code synthesis LM, and up to 4.3{\%} over RL-based models trained with standard PPO or CodeRL. | [
"Gorinski, Philip",
"Zimmer, Matthieu",
"Lampouras, Gerasimos",
"Deik, Derrick Goh Xin",
"Iacobacci, Ignacio"
] | Automatic Unit Test Data Generation and Actor-Critic Reinforcement Learning for Code Synthesis | findings-emnlp.28 | 2310.13669 | [
"https://github.com/huawei-noah/noah-research"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.29.bib | https://aclanthology.org/2023.findings-emnlp.29/ | @inproceedings{leonhardt-etal-2023-unlocking,
title = "Unlocking the Heterogeneous Landscape of Big Data {NLP} with {DUUI}",
author = "Leonhardt, Alexander and
Abrami, Giuseppe and
Baumartz, Daniel and
Mehler, Alexander",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.29",
doi = "10.18653/v1/2023.findings-emnlp.29",
pages = "385--399",
abstract = "Automatic analysis of large corpora is a complex task, especially in terms of time efficiency. This complexity is increased by the fact that flexible, extensible text analysis requires the continuous integration of ever new tools. Since there are no adequate frameworks for these purposes in the field of NLP, and especially in the context of UIMA, that are not outdated or unusable for security reasons, we present a new approach to address the latter task: Docker Unified UIMA Interface (DUUI), a scalable, flexible, lightweight, and feature-rich framework for automatic distributed analysis of text corpora that leverages Big Data experience and virtualization with Docker. We evaluate DUUI{'}s communication approach against a state-of-the-art approach and demonstrate its outstanding behavior in terms of time efficiency, enabling the analysis of big text data.",
}
| Automatic analysis of large corpora is a complex task, especially in terms of time efficiency. This complexity is increased by the fact that flexible, extensible text analysis requires the continuous integration of ever new tools. Since there are no adequate frameworks for these purposes in the field of NLP, and especially in the context of UIMA, that are not outdated or unusable for security reasons, we present a new approach to address the latter task: Docker Unified UIMA Interface (DUUI), a scalable, flexible, lightweight, and feature-rich framework for automatic distributed analysis of text corpora that leverages Big Data experience and virtualization with Docker. We evaluate DUUI{'}s communication approach against a state-of-the-art approach and demonstrate its outstanding behavior in terms of time efficiency, enabling the analysis of big text data. | [
"Leonhardt, Alex",
"er",
"Abrami, Giuseppe",
"Baumartz, Daniel",
"Mehler, Alex",
"er"
] | Unlocking the Heterogeneous Landscape of Big Data NLP with DUUI | findings-emnlp.29 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.30.bib | https://aclanthology.org/2023.findings-emnlp.30/ | @inproceedings{mozes-etal-2023-towards,
title = "Towards Agile Text Classifiers for Everyone",
author = "Mozes, Maximilian and
Hoffmann, Jessica and
Tomanek, Katrin and
Kouate, Muhamed and
Thain, Nithum and
Yuan, Ann and
Bolukbasi, Tolga and
Dixon, Lucas",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.30",
doi = "10.18653/v1/2023.findings-emnlp.30",
pages = "400--414",
abstract = "Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteration and adaptation. This paper introduces and evaluates methods for agile text classification, whereby classifiers are trained using small, targeted datasets that can be quickly developed for a particular policy. Experimenting with 7 datasets from three safety-related domains, comprising 15 annotation schemes, led to our key finding: prompt-tuning large language models, like PaLM 62B, with a labeled dataset of as few as 80 examples can achieve state-of-the-art performance. We argue that this enables a paradigm shift for text classification, especially for models supporting safer online discourse. Instead of collecting millions of examples to attempt to create universal safety classifiers over months or years, classifiers could be tuned using small datasets, created by individuals or small organizations, tailored for specific use cases, and iterated on and adapted in the time-span of a day.",
}
| Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteration and adaptation. This paper introduces and evaluates methods for agile text classification, whereby classifiers are trained using small, targeted datasets that can be quickly developed for a particular policy. Experimenting with 7 datasets from three safety-related domains, comprising 15 annotation schemes, led to our key finding: prompt-tuning large language models, like PaLM 62B, with a labeled dataset of as few as 80 examples can achieve state-of-the-art performance. We argue that this enables a paradigm shift for text classification, especially for models supporting safer online discourse. Instead of collecting millions of examples to attempt to create universal safety classifiers over months or years, classifiers could be tuned using small datasets, created by individuals or small organizations, tailored for specific use cases, and iterated on and adapted in the time-span of a day. | [
"Mozes, Maximilian",
"Hoffmann, Jessica",
"Tomanek, Katrin",
"Kouate, Muhamed",
"Thain, Nithum",
"Yuan, Ann",
"Bolukbasi, Tolga",
"Dixon, Lucas"
] | Towards Agile Text Classifiers for Everyone | findings-emnlp.30 | 2302.06541 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.31.bib | https://aclanthology.org/2023.findings-emnlp.31/ | @inproceedings{adauto-etal-2023-beyond,
title = "Beyond Good Intentions: Reporting the Research Landscape of {NLP} for Social Good",
author = {Adauto, Fernando and
Jin, Zhijing and
Sch{\"o}lkopf, Bernhard and
Hope, Tom and
Sachan, Mrinmaya and
Mihalcea, Rada},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.31",
doi = "10.18653/v1/2023.findings-emnlp.31",
pages = "415--438",
abstract = "With the recent advances in natural language processing (NLP), a vast number of applications have emerged across various use cases. Among the plethora of NLP applications, many academic researchers are motivated to do work that has a positive social impact, in line with the recent initiatives of NLP for Social Good (NLP4SG). However, it is not always obvious to researchers how their research efforts are tackling today{'}s big social problems. Thus, in this paper, we introduce NLP4SGPapers, a scientific dataset with three associated tasks that can help identify NLP4SG papers and characterize the NLP4SG landscape by: (1) identifying the papers that address a social problem, (2) mapping them to the corresponding UN Sustainable Development Goals (SDGs), and (3) identifying the task they are solving and the methods they are using. Using state-of-the-art NLP models, we address each of these tasks and use them on the entire ACL Anthology, resulting in a visualization workspace that gives researchers a comprehensive overview of the field of NLP4SG. Our website is available at https://nlp4sg.vercel.app . We released our data at https://huggingface.co/datasets/feradauto/NLP4SGPapers and code at https://github.com/feradauto/nlp4sg",
}
| With the recent advances in natural language processing (NLP), a vast number of applications have emerged across various use cases. Among the plethora of NLP applications, many academic researchers are motivated to do work that has a positive social impact, in line with the recent initiatives of NLP for Social Good (NLP4SG). However, it is not always obvious to researchers how their research efforts are tackling today{'}s big social problems. Thus, in this paper, we introduce NLP4SGPapers, a scientific dataset with three associated tasks that can help identify NLP4SG papers and characterize the NLP4SG landscape by: (1) identifying the papers that address a social problem, (2) mapping them to the corresponding UN Sustainable Development Goals (SDGs), and (3) identifying the task they are solving and the methods they are using. Using state-of-the-art NLP models, we address each of these tasks and use them on the entire ACL Anthology, resulting in a visualization workspace that gives researchers a comprehensive overview of the field of NLP4SG. Our website is available at https://nlp4sg.vercel.app . We released our data at https://huggingface.co/datasets/feradauto/NLP4SGPapers and code at https://github.com/feradauto/nlp4sg | [
"Adauto, Fern",
"o",
"Jin, Zhijing",
"Sch{\\\"o}lkopf, Bernhard",
"Hope, Tom",
"Sachan, Mrinmaya",
"Mihalcea, Rada"
] | Beyond Good Intentions: Reporting the Research Landscape of NLP for Social Good | findings-emnlp.31 | 2305.05471 | [
"https://github.com/feradauto/nlp4sg"
] | https://huggingface.co/papers/2305.05471 | 1 | 0 | 0 | 7 | [
"feradauto/scibert_nlp4sg"
] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.32.bib | https://aclanthology.org/2023.findings-emnlp.32/ | @inproceedings{li-callison-burch-2023-paxqa,
title = "{PAXQA}: Generating Cross-lingual Question Answering Examples at Training Scale",
author = "Li, Bryan and
Callison-Burch, Chris",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.32",
doi = "10.18653/v1/2023.findings-emnlp.32",
pages = "439--454",
abstract = "Existing question answering (QA) systems owe much of their success to large, high-quality training data. Such annotation efforts are costly, and the difficulty compounds in the cross-lingual setting. Therefore, prior cross-lingual QA work has focused on releasing evaluation datasets, and then applying zero-shot methods as baselines. This work proposes a synthetic data generation method for cross-lingual QA which leverages indirect supervision from existing parallel corpora. Our method termed PAXQA (Projecting annotations for cross-lingual (x) QA) decomposes cross-lingual QA into two stages. First, we apply a question generation (QG) model to the English side. Second, we apply annotation projection to translate both the questions and answers. To better translate questions, we propose a novel use of lexically-constrained machine translation, in which constrained entities are extracted from the parallel bitexts. We apply PAXQA to generate cross-lingual QA examples in 4 languages (662K examples total), and perform human evaluation on a subset to create validation and test splits. We then show that models fine-tuned on these datasets outperform prior synthetic data generation models over several extractive QA datasets. The largest performance gains are for directions with non-English questions and English contexts. Ablation studies show that our dataset generation method is relatively robust to noise from automatic word alignments, showing the sufficient quality of our generations. To facilitate follow-up work, we release our code and datasets at https://github.com/manestay/paxqa.",
}
| Existing question answering (QA) systems owe much of their success to large, high-quality training data. Such annotation efforts are costly, and the difficulty compounds in the cross-lingual setting. Therefore, prior cross-lingual QA work has focused on releasing evaluation datasets, and then applying zero-shot methods as baselines. This work proposes a synthetic data generation method for cross-lingual QA which leverages indirect supervision from existing parallel corpora. Our method termed PAXQA (Projecting annotations for cross-lingual (x) QA) decomposes cross-lingual QA into two stages. First, we apply a question generation (QG) model to the English side. Second, we apply annotation projection to translate both the questions and answers. To better translate questions, we propose a novel use of lexically-constrained machine translation, in which constrained entities are extracted from the parallel bitexts. We apply PAXQA to generate cross-lingual QA examples in 4 languages (662K examples total), and perform human evaluation on a subset to create validation and test splits. We then show that models fine-tuned on these datasets outperform prior synthetic data generation models over several extractive QA datasets. The largest performance gains are for directions with non-English questions and English contexts. Ablation studies show that our dataset generation method is relatively robust to noise from automatic word alignments, showing the sufficient quality of our generations. To facilitate follow-up work, we release our code and datasets at https://github.com/manestay/paxqa. | [
"Li, Bryan",
"Callison-Burch, Chris"
] | PAXQA: Generating Cross-lingual Question Answering Examples at Training Scale | findings-emnlp.32 | 2304.12206 | [
"https://github.com/manestay/paxqa"
] | https://huggingface.co/papers/2304.12206 | 1 | 0 | 0 | 2 | [] | [
"manestay/paxqa_val_test",
"manestay/paxqa_train"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.33.bib | https://aclanthology.org/2023.findings-emnlp.33/ | @inproceedings{cao-etal-2023-sharing,
title = "Sharing, Teaching and Aligning: Knowledgeable Transfer Learning for Cross-Lingual Machine Reading Comprehension",
author = "Cao, Tingfeng and
Wang, Chengyu and
Tan, Chuanqi and
Huang, Jun and
Zhu, Jinhui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.33",
doi = "10.18653/v1/2023.findings-emnlp.33",
pages = "455--467",
abstract = "In cross-lingual language understanding, machine translation is often utilized to enhance the transferability of models across languages, either by translating the training data from the source language to the target, or from the target to the source to aid inference. However, in cross-lingual machine reading comprehension (MRC), it is difficult to perform a deep level of assistance to enhance cross-lingual transfer because of the variation of answer span positions in different languages. In this paper, we propose X-STA, a new approach for cross-lingual MRC. Specifically, we leverage an attentive teacher to subtly transfer the answer spans of the source language to the answer output space of the target. A Gradient-Disentangled Knowledge Sharing technique is proposed as an improved cross-attention block. In addition, we force the model to learn semantic alignments from multiple granularities and calibrate the model outputs with teacher guidance to enhance cross-lingual transferability. Experiments on three multi-lingual MRC datasets show the effectiveness of our method, outperforming state-of-the-art approaches.",
}
| In cross-lingual language understanding, machine translation is often utilized to enhance the transferability of models across languages, either by translating the training data from the source language to the target, or from the target to the source to aid inference. However, in cross-lingual machine reading comprehension (MRC), it is difficult to perform a deep level of assistance to enhance cross-lingual transfer because of the variation of answer span positions in different languages. In this paper, we propose X-STA, a new approach for cross-lingual MRC. Specifically, we leverage an attentive teacher to subtly transfer the answer spans of the source language to the answer output space of the target. A Gradient-Disentangled Knowledge Sharing technique is proposed as an improved cross-attention block. In addition, we force the model to learn semantic alignments from multiple granularities and calibrate the model outputs with teacher guidance to enhance cross-lingual transferability. Experiments on three multi-lingual MRC datasets show the effectiveness of our method, outperforming state-of-the-art approaches. | [
"Cao, Tingfeng",
"Wang, Chengyu",
"Tan, Chuanqi",
"Huang, Jun",
"Zhu, Jinhui"
] | Sharing, Teaching and Aligning: Knowledgeable Transfer Learning for Cross-Lingual Machine Reading Comprehension | findings-emnlp.33 | 2311.06758 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.34.bib | https://aclanthology.org/2023.findings-emnlp.34/ | @inproceedings{roussinov-sharoff-2023-bert,
title = "{BERT} Goes Off-Topic: Investigating the Domain Transfer Challenge using Genre Classification",
author = "Roussinov, Dmitri and
Sharoff, Serge",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.34",
doi = "10.18653/v1/2023.findings-emnlp.34",
pages = "468--483",
abstract = "While performance of many text classification tasks has been recently improved due to Pretrained Language Models (PLMs), in this paper we show that they still suffer from a performance gap when the underlying distribution of topics changes. For example, a genre classifier trained on political topics often fails when tested on documents in the same genre, but about sport or medicine. In this work, we quantify this phenomenon empirically with a large corpus and a large set of topics. Thus, we verify that domain transfer remains challenging both for classic PLMs, such as BERT, and for modern large models (LLMs), such as GPT. We develop a data augmentation approach by generating texts in any desired genre and on any desired topic, even when there are no documents in the training corpus that are both in that particular genre and on that particular topic. When we augment the training dataset with the topically-controlled synthetic texts, F1 improves up to 50{\%} for some topics, approaching on-topic training, while showing no or next to no improvement for other topics. While our empirical results focus on genre classification, our methodology is applicable to other classification tasks such as gender, authorship, or sentiment classification.",
}
| While performance of many text classification tasks has been recently improved due to Pretrained Language Models (PLMs), in this paper we show that they still suffer from a performance gap when the underlying distribution of topics changes. For example, a genre classifier trained on political topics often fails when tested on documents in the same genre, but about sport or medicine. In this work, we quantify this phenomenon empirically with a large corpus and a large set of topics. Thus, we verify that domain transfer remains challenging both for classic PLMs, such as BERT, and for modern large models (LLMs), such as GPT. We develop a data augmentation approach by generating texts in any desired genre and on any desired topic, even when there are no documents in the training corpus that are both in that particular genre and on that particular topic. When we augment the training dataset with the topically-controlled synthetic texts, F1 improves up to 50{\%} for some topics, approaching on-topic training, while showing no or next to no improvement for other topics. While our empirical results focus on genre classification, our methodology is applicable to other classification tasks such as gender, authorship, or sentiment classification. | [
"Roussinov, Dmitri",
"Sharoff, Serge"
] | BERT Goes Off-Topic: Investigating the Domain Transfer Challenge using Genre Classification | findings-emnlp.34 | 2311.16083 | [
"https://github.com/dminus1/genre"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.35.bib | https://aclanthology.org/2023.findings-emnlp.35/ | @inproceedings{colombo-etal-2023-toward,
title = "Toward Stronger Textual Attack Detectors",
author = "Colombo, Pierre and
Picot, Marine and
Noiry, Nathan and
Staerman, Guillaume and
Piantanida, Pablo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.35",
doi = "10.18653/v1/2023.findings-emnlp.35",
pages = "484--505",
abstract = "The landscape of available textual adversarial attacks keeps growing, posing severe threats and raising concerns regarding deep NLP systems integrity. However, the crucial problem of defending against malicious attacks has only drawn few attention in the NLP community. The latter is nonetheless instrumental to develop robust and trustworthy systems. This paper makes two important contributions in this line of search: \textit{(i)} we introduce LAROUSSE, a new framework to detect textual adversarial attacks and \textit{(ii)} we introduce STAKEOUT, an extended benchmark composed of nine popular attack methods, three datasets and two pre-trained models. LAROUSSE is ready-to-use in production as it is unsupervised, hyperparameter free and non-differentiable, protecting it against gradient-based methods. Our new benchmark STAKEOUT allows for a robust evaluation framework: we conduct extensive numerical experiments which demonstrate that LAROUSSE outperforms previous methods, and which allows to identify interesting factor of detection rate variations.",
}
| The landscape of available textual adversarial attacks keeps growing, posing severe threats and raising concerns regarding deep NLP systems integrity. However, the crucial problem of defending against malicious attacks has only drawn few attention in the NLP community. The latter is nonetheless instrumental to develop robust and trustworthy systems. This paper makes two important contributions in this line of search: \textit{(i)} we introduce LAROUSSE, a new framework to detect textual adversarial attacks and \textit{(ii)} we introduce STAKEOUT, an extended benchmark composed of nine popular attack methods, three datasets and two pre-trained models. LAROUSSE is ready-to-use in production as it is unsupervised, hyperparameter free and non-differentiable, protecting it against gradient-based methods. Our new benchmark STAKEOUT allows for a robust evaluation framework: we conduct extensive numerical experiments which demonstrate that LAROUSSE outperforms previous methods, and which allows to identify interesting factor of detection rate variations. | [
"Colombo, Pierre",
"Picot, Marine",
"Noiry, Nathan",
"Staerman, Guillaume",
"Piantanida, Pablo"
] | Toward Stronger Textual Attack Detectors | findings-emnlp.35 | 2310.14001 | [
"https://github.com/pierrecolombo/adversarialattacksnlp"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.36.bib | https://aclanthology.org/2023.findings-emnlp.36/ | @inproceedings{koksal-etal-2023-meal,
title = "{MEAL}: Stable and Active Learning for Few-Shot Prompting",
author = {K{\"o}ksal, Abdullatif and
Schick, Timo and
Schuetze, Hinrich},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.36",
doi = "10.18653/v1/2023.findings-emnlp.36",
pages = "506--517",
abstract = "Few-shot classification has made great strides due to foundation models that, through priming and prompting, are highly effective few-shot learners. However, this approach has high variance both across different sets of few shots (*data selection*) and across different finetuning runs (*run variability*). This is problematic not only because it impedes the fair comparison of different approaches, but especially because it makes few-shot learning too unreliable for many real-world applications. To alleviate these issues, we make two contributions for more stable and effective few-shot learning: First, we propose novel ensembling methods and show that they substantially reduce *run variability*. Second, we introduce a new active learning (AL) criterion for *data selection* and present the first AL-based approach specifically tailored towards prompt-based learning. In our experiments, we show that our combined method, MEAL (**M**ultiprompt finetuning and prediction **E**nsembling with **A**ctive **L**earning), improves overall performance of prompt-based finetuning by 2.3 points on five diverse tasks. We publicly share our code and data splits in https://github.com/akoksal/MEAL.",
}
| Few-shot classification has made great strides due to foundation models that, through priming and prompting, are highly effective few-shot learners. However, this approach has high variance both across different sets of few shots (*data selection*) and across different finetuning runs (*run variability*). This is problematic not only because it impedes the fair comparison of different approaches, but especially because it makes few-shot learning too unreliable for many real-world applications. To alleviate these issues, we make two contributions for more stable and effective few-shot learning: First, we propose novel ensembling methods and show that they substantially reduce *run variability*. Second, we introduce a new active learning (AL) criterion for *data selection* and present the first AL-based approach specifically tailored towards prompt-based learning. In our experiments, we show that our combined method, MEAL (**M**ultiprompt finetuning and prediction **E**nsembling with **A**ctive **L**earning), improves overall performance of prompt-based finetuning by 2.3 points on five diverse tasks. We publicly share our code and data splits in https://github.com/akoksal/MEAL. | [
"K{\\\"o}ksal, Abdullatif",
"Schick, Timo",
"Schuetze, Hinrich"
] | MEAL: Stable and Active Learning for Few-Shot Prompting | findings-emnlp.36 | 2211.08358 | [
"https://github.com/akoksal/meal"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.37.bib | https://aclanthology.org/2023.findings-emnlp.37/ | @inproceedings{zhang-etal-2023-structure,
title = "Structure and Label Constrained Data Augmentation for Cross-domain Few-shot {NER}",
author = "Zhang, Jingyi and
Zhang, Ying and
Chen, Yufeng and
Xu, Jinan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.37",
doi = "10.18653/v1/2023.findings-emnlp.37",
pages = "518--530",
abstract = "Cross-domain few-shot named entity recognition (NER) is a challenging task that aims to recognize entities in target domains with limited labeled data by leveraging relevant knowledge from source domains. However, domain gaps limit the effect of knowledge transfer and harm the performance of NER models. In this paper, we analyze those domain gaps from two new perspectives, i.e., entity annotations and entity structures and leverage word-to-tag and word-to-word relations to model them, respectively. Moreover, we propose a novel method called Structure and Label Constrained Data Augmentation (SLC-DA) for Cross-domain Few-shot NER, which novelly design a label constrained pre-train task and a structure constrained optimization objectives in the data augmentation process to generate domain-specific augmented data to help NER models smoothly transition from source to target domains. We evaluate our approach on several standard datasets and achieve state-of-the-art or competitive results, demonstrating the effectiveness of our method in cross-domain few-shot NER.",
}
| Cross-domain few-shot named entity recognition (NER) is a challenging task that aims to recognize entities in target domains with limited labeled data by leveraging relevant knowledge from source domains. However, domain gaps limit the effect of knowledge transfer and harm the performance of NER models. In this paper, we analyze those domain gaps from two new perspectives, i.e., entity annotations and entity structures and leverage word-to-tag and word-to-word relations to model them, respectively. Moreover, we propose a novel method called Structure and Label Constrained Data Augmentation (SLC-DA) for Cross-domain Few-shot NER, which novelly design a label constrained pre-train task and a structure constrained optimization objectives in the data augmentation process to generate domain-specific augmented data to help NER models smoothly transition from source to target domains. We evaluate our approach on several standard datasets and achieve state-of-the-art or competitive results, demonstrating the effectiveness of our method in cross-domain few-shot NER. | [
"Zhang, Jingyi",
"Zhang, Ying",
"Chen, Yufeng",
"Xu, Jinan"
] | Structure and Label Constrained Data Augmentation for Cross-domain Few-shot NER | findings-emnlp.37 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.38.bib | https://aclanthology.org/2023.findings-emnlp.38/ | @inproceedings{goswami-etal-2023-weakly,
title = "Weakly-supervised Deep Cognate Detection Framework for Low-Resourced Languages Using Morphological Knowledge of Closely-Related Languages",
author = "Goswami, Koustava and
Rani, Priya and
Fransen, Theodorus and
McCrae, John",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.38",
doi = "10.18653/v1/2023.findings-emnlp.38",
pages = "531--541",
abstract = "Exploiting cognates for transfer learning in under-resourced languages is an exciting opportunity for language understanding tasks, including unsupervised machine translation, named entity recognition and information retrieval. Previous approaches mainly focused on supervised cognate detection tasks based on orthographic, phonetic or state-of-the-art contextual language models, which under-perform for most under-resourced languages. This paper proposes a novel language-agnostic weakly-supervised deep cognate detection framework for under-resourced languages using morphological knowledge from closely related languages. We train an encoder to gain morphological knowledge of a language and transfer the knowledge to perform unsupervised and weakly-supervised cognate detection tasks with and without the pivot language for the closely-related languages. While unsupervised, it overcomes the need for hand-crafted annotation of cognates. We performed experiments on different published cognate detection datasets across language families and observed not only significant improvement over the state-of-the-art but also our method outperformed the state-of-the-art supervised and unsupervised methods. Our model can be extended to a wide range of languages from any language family as it overcomes the requirement of the annotation of the cognate pairs for training.",
}
| Exploiting cognates for transfer learning in under-resourced languages is an exciting opportunity for language understanding tasks, including unsupervised machine translation, named entity recognition and information retrieval. Previous approaches mainly focused on supervised cognate detection tasks based on orthographic, phonetic or state-of-the-art contextual language models, which under-perform for most under-resourced languages. This paper proposes a novel language-agnostic weakly-supervised deep cognate detection framework for under-resourced languages using morphological knowledge from closely related languages. We train an encoder to gain morphological knowledge of a language and transfer the knowledge to perform unsupervised and weakly-supervised cognate detection tasks with and without the pivot language for the closely-related languages. While unsupervised, it overcomes the need for hand-crafted annotation of cognates. We performed experiments on different published cognate detection datasets across language families and observed not only significant improvement over the state-of-the-art but also our method outperformed the state-of-the-art supervised and unsupervised methods. Our model can be extended to a wide range of languages from any language family as it overcomes the requirement of the annotation of the cognate pairs for training. | [
"Goswami, Koustava",
"Rani, Priya",
"Fransen, Theodorus",
"McCrae, John"
] | Weakly-supervised Deep Cognate Detection Framework for Low-Resourced Languages Using Morphological Knowledge of Closely-Related Languages | findings-emnlp.38 | 2311.05155 | [
"https://github.com/koustavagoswami/weakly_supervised-cognate_detection"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.39.bib | https://aclanthology.org/2023.findings-emnlp.39/ | @inproceedings{sun-etal-2023-sqlprompt,
title = "{SQLP}rompt: In-Context Text-to-{SQL} with Minimal Labeled Data",
author = "Sun, Ruoxi and
Arik, Sercan and
Sinha, Rajarishi and
Nakhost, Hootan and
Dai, Hanjun and
Yin, Pengcheng and
Pfister, Tomas",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.39",
doi = "10.18653/v1/2023.findings-emnlp.39",
pages = "542--550",
abstract = "Text-to-SQL aims to automate the process of generating SQL queries on a database from natural language text. In this work, we propose {``}SQLPrompt{''}, tailored to improve the few-shot prompting capabilities of Text-to-SQL for Large Language Models (LLMs). Our methods include innovative prompt design, execution-based consistency decoding strategy which selects the SQL with the most consistent execution outcome among other SQL proposals, and a method that aims to improve performance by diversifying the SQL proposals during consistency selection with different prompt designs ({``}MixPrompt{''}) and foundation models ({``}MixLLMs{''}). We show that \textit{SQLPrompt} outperforms previous approaches for in-context learning with zero labeled data by a large margin, closing the gap with finetuning state-of-the-art with thousands of labeled data.",
}
| Text-to-SQL aims to automate the process of generating SQL queries on a database from natural language text. In this work, we propose {``}SQLPrompt{''}, tailored to improve the few-shot prompting capabilities of Text-to-SQL for Large Language Models (LLMs). Our methods include innovative prompt design, execution-based consistency decoding strategy which selects the SQL with the most consistent execution outcome among other SQL proposals, and a method that aims to improve performance by diversifying the SQL proposals during consistency selection with different prompt designs ({``}MixPrompt{''}) and foundation models ({``}MixLLMs{''}). We show that \textit{SQLPrompt} outperforms previous approaches for in-context learning with zero labeled data by a large margin, closing the gap with finetuning state-of-the-art with thousands of labeled data. | [
"Sun, Ruoxi",
"Arik, Sercan",
"Sinha, Rajarishi",
"Nakhost, Hootan",
"Dai, Hanjun",
"Yin, Pengcheng",
"Pfister, Tomas"
] | SQLPrompt: In-Context Text-to-SQL with Minimal Labeled Data | findings-emnlp.39 | 2311.02883 | [
""
] | https://huggingface.co/papers/2311.02883 | 0 | 0 | 0 | 7 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.40.bib | https://aclanthology.org/2023.findings-emnlp.40/ | @inproceedings{zhang-etal-2023-toward,
title = "Toward Building General Foundation Models for Language, Vision, and Vision-Language Understanding Tasks",
author = "Zhang, Xinsong and
Zeng, Yan and
Zhang, Jipeng and
Li, Hang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.40",
doi = "10.18653/v1/2023.findings-emnlp.40",
pages = "551--568",
abstract = "Foundation models or pre-trained models have substantially improved the performance of various language, vision, and vision-language understanding tasks. However, existing foundation models can only perform the best in one type of tasks, namely language, vision, or vision-language. It is still an open question whether it is possible to construct a general foundation model performing the best for all the understanding tasks. In this paper, we propose a new method for training the general foundation model, X-FM (the X-Foundation Model). X-FM has one language encoder, one vision encoder, and one fusion encoder, as well as a new training method. The training method includes two new techniques for learning X-FM from text, image, and image-text pair data. One is to stop gradients from the vision-language training when learning the language encoder. The other is to leverage the vision-language training to guide the learning of the vision encoder. Extensive experiments on benchmark datasets show that X-FM can significantly outperform existing general foundation models and perform better than or comparable to existing foundation models specifically for language, vision, or vision-language understanding. Code and pre-trained models are released at https://github.com/zhangxinsong-nlp/XFM.",
}
| Foundation models or pre-trained models have substantially improved the performance of various language, vision, and vision-language understanding tasks. However, existing foundation models can only perform the best in one type of tasks, namely language, vision, or vision-language. It is still an open question whether it is possible to construct a general foundation model performing the best for all the understanding tasks. In this paper, we propose a new method for training the general foundation model, X-FM (the X-Foundation Model). X-FM has one language encoder, one vision encoder, and one fusion encoder, as well as a new training method. The training method includes two new techniques for learning X-FM from text, image, and image-text pair data. One is to stop gradients from the vision-language training when learning the language encoder. The other is to leverage the vision-language training to guide the learning of the vision encoder. Extensive experiments on benchmark datasets show that X-FM can significantly outperform existing general foundation models and perform better than or comparable to existing foundation models specifically for language, vision, or vision-language understanding. Code and pre-trained models are released at https://github.com/zhangxinsong-nlp/XFM. | [
"Zhang, Xinsong",
"Zeng, Yan",
"Zhang, Jipeng",
"Li, Hang"
] | Toward Building General Foundation Models for Language, Vision, and Vision-Language Understanding Tasks | findings-emnlp.40 | 2301.05065 | [
"https://github.com/zhangxinsong-nlp/XFM"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.41.bib | https://aclanthology.org/2023.findings-emnlp.41/ | @inproceedings{wolska-etal-2023-trigger,
title = "Trigger Warnings: Bootstrapping a Violence Detector for Fan Fiction",
author = {Wolska, Magdalena and
Wiegmann, Matti and
Schr{\"o}der, Christopher and
Borchardt, Ole and
Stein, Benno and
Potthast, Martin},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.41",
doi = "10.18653/v1/2023.findings-emnlp.41",
pages = "569--576",
abstract = "We present the first dataset and evaluation results on a newly defined task: assigning trigger warnings. We introduce a labeled corpus of narrative fiction from Archive of Our Own (AO3), a popular fan fiction site, and define a document-level classification task to determine whether or not to assign a trigger warning to an English story. We focus on the most commonly assigned trigger type {``}violence{'} using the warning labels provided by AO3 authors as ground-truth labels. We trained SVM, BERT, and Longfomer models on three datasets sampled from the corpus and achieve F1 scores between 0.8 and 0.9, indicating that assigning trigger warnings for violence is feasible.",
}
| We present the first dataset and evaluation results on a newly defined task: assigning trigger warnings. We introduce a labeled corpus of narrative fiction from Archive of Our Own (AO3), a popular fan fiction site, and define a document-level classification task to determine whether or not to assign a trigger warning to an English story. We focus on the most commonly assigned trigger type {``}violence{'} using the warning labels provided by AO3 authors as ground-truth labels. We trained SVM, BERT, and Longfomer models on three datasets sampled from the corpus and achieve F1 scores between 0.8 and 0.9, indicating that assigning trigger warnings for violence is feasible. | [
"Wolska, Magdalena",
"Wiegmann, Matti",
"Schr{\\\"o}der, Christopher",
"Borchardt, Ole",
"Stein, Benno",
"Potthast, Martin"
] | Trigger Warnings: Bootstrapping a Violence Detector for Fan Fiction | findings-emnlp.41 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.42.bib | https://aclanthology.org/2023.findings-emnlp.42/ | @inproceedings{chen-etal-2023-pass,
title = "Pass-Tuning: Towards Structure-Aware Parameter-Efficient Tuning for Code Representation Learning",
author = "Chen, Nuo and
Sun, Qiushi and
Wang, Jianing and
Li, Xiang and
Gao, Ming",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.42",
doi = "10.18653/v1/2023.findings-emnlp.42",
pages = "577--591",
abstract = "Code pre-trained models (CodePTMs) have recently become the de-facto paradigm for various tasks in the domain of code intelligence. To achieve excellent performance, the widely used strategy is to fine-tune all the parameters of CodePTMs. However, as the model size increases along with the number of downstream tasks, this strategy becomes excessively expensive. There are also some prior works that utilize Parameter-Efficient Learning (PEL) methods for model tuning in natural language processing to mitigate similar problems, but applying them directly to CodePTMs fails to capture the inherent structural characteristics of codes. To address the problem, in this paper, we propose Pass-Tuning for structure-aware Parameter-Efficient code representation learning. Specifically, a plug-and-play graph neural network module that can learn from Abstract Syntax Tree (AST) is employed as a tunable prefix. On the one hand, Pass-Tuning can further exploit the structural information of source code. On the other hand, it could serve as a replacement for full fine-tuning. We evaluate our method on multiple tasks across eight programming languages, including code understanding and generation. These results demonstrate the effectiveness, robustness, and universality of our method.",
}
| Code pre-trained models (CodePTMs) have recently become the de-facto paradigm for various tasks in the domain of code intelligence. To achieve excellent performance, the widely used strategy is to fine-tune all the parameters of CodePTMs. However, as the model size increases along with the number of downstream tasks, this strategy becomes excessively expensive. There are also some prior works that utilize Parameter-Efficient Learning (PEL) methods for model tuning in natural language processing to mitigate similar problems, but applying them directly to CodePTMs fails to capture the inherent structural characteristics of codes. To address the problem, in this paper, we propose Pass-Tuning for structure-aware Parameter-Efficient code representation learning. Specifically, a plug-and-play graph neural network module that can learn from Abstract Syntax Tree (AST) is employed as a tunable prefix. On the one hand, Pass-Tuning can further exploit the structural information of source code. On the other hand, it could serve as a replacement for full fine-tuning. We evaluate our method on multiple tasks across eight programming languages, including code understanding and generation. These results demonstrate the effectiveness, robustness, and universality of our method. | [
"Chen, Nuo",
"Sun, Qiushi",
"Wang, Jianing",
"Li, Xiang",
"Gao, Ming"
] | Pass-Tuning: Towards Structure-Aware Parameter-Efficient Tuning for Code Representation Learning | findings-emnlp.42 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.43.bib | https://aclanthology.org/2023.findings-emnlp.43/ | @inproceedings{lin-etal-2023-counterfactual,
title = "Counterfactual Augmentation for Multimodal Learning Under Presentation Bias",
author = "Lin, Victoria and
Morency, Louis-Philippe and
Dimitriadis, Dimitrios and
Sharma, Srinagesh",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.43",
doi = "10.18653/v1/2023.findings-emnlp.43",
pages = "592--606",
abstract = "In real-world machine learning systems, labels are often derived from user behaviors that the system wishes to encourage. Over time, new models must be trained as new training examples and features become available. However, feedback loops between users and models can bias future user behavior, inducing a *presentation bias* in the labels that compromises the ability to train new models. In this paper, we propose *counterfactual augmentation*, a novel causal method for correcting presentation bias using generated counterfactual labels. Our empirical evaluations demonstrate that counterfactual augmentation yields better downstream performance compared to both uncorrected models and existing bias-correction methods. Model analyses further indicate that the generated counterfactuals align closely with true counterfactuals in an oracle setting.",
}
| In real-world machine learning systems, labels are often derived from user behaviors that the system wishes to encourage. Over time, new models must be trained as new training examples and features become available. However, feedback loops between users and models can bias future user behavior, inducing a *presentation bias* in the labels that compromises the ability to train new models. In this paper, we propose *counterfactual augmentation*, a novel causal method for correcting presentation bias using generated counterfactual labels. Our empirical evaluations demonstrate that counterfactual augmentation yields better downstream performance compared to both uncorrected models and existing bias-correction methods. Model analyses further indicate that the generated counterfactuals align closely with true counterfactuals in an oracle setting. | [
"Lin, Victoria",
"Morency, Louis-Philippe",
"Dimitriadis, Dimitrios",
"Sharma, Srinagesh"
] | Counterfactual Augmentation for Multimodal Learning Under Presentation Bias | findings-emnlp.43 | 2305.14083 | [
"https://github.com/microsoft/causaltransfer"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.44.bib | https://aclanthology.org/2023.findings-emnlp.44/ | @inproceedings{chen-etal-2023-table,
title = "A Table-to-Text Framework with Heterogeneous Multidominance Attention and Self-Evaluated Multi-Pass Deliberation",
author = "Chen, Xi and
Lu, Xinjiang and
Xin, Haoran and
Peng, Wenjun and
Duan, Haoyang and
Jiang, Feihu and
Zhou, Jingbo and
Xiong, Hui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.44",
doi = "10.18653/v1/2023.findings-emnlp.44",
pages = "607--620",
abstract = "Though big progress in table-to-text works, effectively leveraging table structure signals, e.g., hierarchical structure, remains challenging. Besides, deliberating generated descriptions proves to be effective for table-to-text. However, determining the appropriate outcome when encountering multi-pass candidates is another challenge. To this end, we propose a novel table-to-text approach on top of Self-evaluated multi-pass Generation and Heterogenous Multidominance Attention, namely SG-HMA. Specifically, we formulate the table structure into a multidominance (MD) structure and devise a heterogenous multidominance attention (HMA) to comprehensively explore the complex interactions encoded in the hierarchical structure, which can further deliver rich signals for text generation with the help of pre-trained language models (PLMs). Afterward, a contrastive loss is introduced to align the generation objective with evaluation metrics, so the more faithful generated descriptions can be guaranteed. We conduct extensive experiments on three public datasets, demonstrating that SG-HMA outperforms several SOTA methods quantitatively and qualitatively.",
}
| Though big progress in table-to-text works, effectively leveraging table structure signals, e.g., hierarchical structure, remains challenging. Besides, deliberating generated descriptions proves to be effective for table-to-text. However, determining the appropriate outcome when encountering multi-pass candidates is another challenge. To this end, we propose a novel table-to-text approach on top of Self-evaluated multi-pass Generation and Heterogenous Multidominance Attention, namely SG-HMA. Specifically, we formulate the table structure into a multidominance (MD) structure and devise a heterogenous multidominance attention (HMA) to comprehensively explore the complex interactions encoded in the hierarchical structure, which can further deliver rich signals for text generation with the help of pre-trained language models (PLMs). Afterward, a contrastive loss is introduced to align the generation objective with evaluation metrics, so the more faithful generated descriptions can be guaranteed. We conduct extensive experiments on three public datasets, demonstrating that SG-HMA outperforms several SOTA methods quantitatively and qualitatively. | [
"Chen, Xi",
"Lu, Xinjiang",
"Xin, Haoran",
"Peng, Wenjun",
"Duan, Haoyang",
"Jiang, Feihu",
"Zhou, Jingbo",
"Xiong, Hui"
] | A Table-to-Text Framework with Heterogeneous Multidominance Attention and Self-Evaluated Multi-Pass Deliberation | findings-emnlp.44 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.45.bib | https://aclanthology.org/2023.findings-emnlp.45/ | @inproceedings{zou-etal-2023-crossing,
title = "Crossing the Aisle: Unveiling Partisan and Counter-Partisan Events in News Reporting",
author = "Zou, Kaijian and
Zhang, Xinliang and
Wu, Winston and
Beauchamp, Nicholas and
Wang, Lu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.45",
doi = "10.18653/v1/2023.findings-emnlp.45",
pages = "621--632",
abstract = "News media is expected to uphold unbiased reporting. Yet they may still affect public opinion by selectively including or omitting events that support or contradict their ideological positions. Prior work in NLP has only studied media bias via linguistic style and word usage. In this paper, we study to which degree media balances news reporting and affects consumers through event inclusion or omission. We first introduce the task of detecting both partisan and counter-partisan events: events that support or oppose the author{'}s political ideology. To conduct our study, we annotate a high-quality dataset, PAC, containing 8,511 (counter-)partisan event annotations in 304 news articles from ideologically diverse media outlets. We benchmark PAC to highlight the challenges of this task. Our findings highlight both the ways in which the news subtly shapes opinion and the need for large language models that better understand events within a broader context. Our dataset can be found at https://github.com/launchnlp/Partisan-Event-Dataset.",
}
| News media is expected to uphold unbiased reporting. Yet they may still affect public opinion by selectively including or omitting events that support or contradict their ideological positions. Prior work in NLP has only studied media bias via linguistic style and word usage. In this paper, we study to which degree media balances news reporting and affects consumers through event inclusion or omission. We first introduce the task of detecting both partisan and counter-partisan events: events that support or oppose the author{'}s political ideology. To conduct our study, we annotate a high-quality dataset, PAC, containing 8,511 (counter-)partisan event annotations in 304 news articles from ideologically diverse media outlets. We benchmark PAC to highlight the challenges of this task. Our findings highlight both the ways in which the news subtly shapes opinion and the need for large language models that better understand events within a broader context. Our dataset can be found at https://github.com/launchnlp/Partisan-Event-Dataset. | [
"Zou, Kaijian",
"Zhang, Xinliang",
"Wu, Winston",
"Beauchamp, Nicholas",
"Wang, Lu"
] | Crossing the Aisle: Unveiling Partisan and Counter-Partisan Events in News Reporting | findings-emnlp.45 | 2310.18768 | [
"https://github.com/launchnlp/partisan-event-dataset"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.46.bib | https://aclanthology.org/2023.findings-emnlp.46/ | @inproceedings{wang-shi-2023-video,
title = "Video-Text Retrieval by Supervised Sparse Multi-Grained Learning",
author = "Wang, Yimu and
Shi, Peng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.46",
doi = "10.18653/v1/2023.findings-emnlp.46",
pages = "633--649",
abstract = "While recent progress in video-text retrieval has been advanced by the exploration of better representation learning, in this paper, we present a novel multi-grained sparse learning framework, S3MA, to learn an aligned sparse space shared between the video and the text for video-text retrieval. The shared sparse space is initialized with a finite number of sparse concepts, each of which refers to a number of words. With the text data at hand, we learn and update the shared sparse space in a supervised manner using the proposed similarity and alignment losses. Moreover, to enable multi-grained alignment, we incorporate frame representations for better modeling the video modality and calculating fine-grained and coarse-grained similarities. Benefiting from the learned shared sparse space and multi-grained similarities, extensive experiments on several video-text retrieval benchmarks demonstrate the superiority of S3MA over existing methods.",
}
| While recent progress in video-text retrieval has been advanced by the exploration of better representation learning, in this paper, we present a novel multi-grained sparse learning framework, S3MA, to learn an aligned sparse space shared between the video and the text for video-text retrieval. The shared sparse space is initialized with a finite number of sparse concepts, each of which refers to a number of words. With the text data at hand, we learn and update the shared sparse space in a supervised manner using the proposed similarity and alignment losses. Moreover, to enable multi-grained alignment, we incorporate frame representations for better modeling the video modality and calculating fine-grained and coarse-grained similarities. Benefiting from the learned shared sparse space and multi-grained similarities, extensive experiments on several video-text retrieval benchmarks demonstrate the superiority of S3MA over existing methods. | [
"Wang, Yimu",
"Shi, Peng"
] | Video-Text Retrieval by Supervised Sparse Multi-Grained Learning | findings-emnlp.46 | 2302.09473 | [
"https://github.com/yimuwangcs/Better_Cross_Modal_Retrieval"
] | https://huggingface.co/papers/2302.09473 | 1 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.47.bib | https://aclanthology.org/2023.findings-emnlp.47/ | @inproceedings{comi-etal-2023-zero,
title = "Zero-Shot-{BERT}-Adapters: a Zero-Shot Pipeline for Unknown Intent Detection",
author = "Comi, Daniele and
Christofidellis, Dimitrios and
Piazza, Pier and
Manica, Matteo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.47",
doi = "10.18653/v1/2023.findings-emnlp.47",
pages = "650--663",
abstract = "Intent discovery is a crucial task in natural language processing, and it is increasingly relevant for various of industrial applications. Identifying novel, unseen intents from user inputs remains one of the biggest challenges in this field. Herein, we propose Zero-Shot-BERT-Adapters, a two-stage method for multilingual intent discovery relying on a Transformer architecture, fine-tuned with Adapters. We train the model for Natural Language Inference (NLI) and later perform unknown intent classification in a zero-shot setting for multiple languages. In our evaluation, we first analyze the quality of the model after adaptive fine-tuning on known classes. Secondly, we evaluate its performance in casting intent classification as an NLI task. Lastly, we test the zero-shot performance of the model on unseen classes, showing how Zero-Shot-BERT-Adapters can effectively perform intent discovery by generating semantically similar intents, if not equal, to the ground-truth ones. Our experiments show how Zero-Shot-BERT-Adapters outperforms various baselines in two zero-shot settings: known intent classification and unseen intent discovery. The proposed pipeline holds the potential for broad application in customer care. It enables automated dynamic triage using a lightweight model that can be easily deployed and scaled in various business scenarios, unlike large language models. Zero-Shot-BERT-Adapters represents an innovative multi-language approach for intent discovery, enabling the online generation of novel intents. A Python package implementing the pipeline and the new datasets we compiled are available at the following link: https://github.com/GT4SD/zero-shot-bert-adapters.",
}
| Intent discovery is a crucial task in natural language processing, and it is increasingly relevant for various of industrial applications. Identifying novel, unseen intents from user inputs remains one of the biggest challenges in this field. Herein, we propose Zero-Shot-BERT-Adapters, a two-stage method for multilingual intent discovery relying on a Transformer architecture, fine-tuned with Adapters. We train the model for Natural Language Inference (NLI) and later perform unknown intent classification in a zero-shot setting for multiple languages. In our evaluation, we first analyze the quality of the model after adaptive fine-tuning on known classes. Secondly, we evaluate its performance in casting intent classification as an NLI task. Lastly, we test the zero-shot performance of the model on unseen classes, showing how Zero-Shot-BERT-Adapters can effectively perform intent discovery by generating semantically similar intents, if not equal, to the ground-truth ones. Our experiments show how Zero-Shot-BERT-Adapters outperforms various baselines in two zero-shot settings: known intent classification and unseen intent discovery. The proposed pipeline holds the potential for broad application in customer care. It enables automated dynamic triage using a lightweight model that can be easily deployed and scaled in various business scenarios, unlike large language models. Zero-Shot-BERT-Adapters represents an innovative multi-language approach for intent discovery, enabling the online generation of novel intents. A Python package implementing the pipeline and the new datasets we compiled are available at the following link: https://github.com/GT4SD/zero-shot-bert-adapters. | [
"Comi, Daniele",
"Christofidellis, Dimitrios",
"Piazza, Pier",
"Manica, Matteo"
] | Zero-Shot-BERT-Adapters: a Zero-Shot Pipeline for Unknown Intent Detection | findings-emnlp.47 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.48.bib | https://aclanthology.org/2023.findings-emnlp.48/ | @inproceedings{zhang-etal-2023-refsql,
title = "{R}e{FSQL}: A Retrieval-Augmentation Framework for Text-to-{SQL} Generation",
author = "Zhang, Kun and
Lin, Xiexiong and
Wang, Yuanzhuo and
Zhang, Xin and
Sun, Fei and
Jianhe, Cen and
Tan, Hexiang and
Jiang, Xuhui and
Shen, Huawei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.48",
doi = "10.18653/v1/2023.findings-emnlp.48",
pages = "664--673",
abstract = "Text-to-SQL is the task that aims at translating natural language questions into SQL queries. Existing methods directly align the natural language with SQL Language and train one encoder-decoder-based model to fit all questions. However, they underestimate the inherent structural characteristics of SQL, as well as the gap between specific structure knowledge and general knowledge. This leads to structure errors in the generated SQL. To address the above challenges, we propose a retrieval-argument framework, namely ReFSQL. It contains two parts, structure-enhanced retriever and the generator. Structure-enhanced retriever is designed to identify samples with comparable specific knowledge in an unsupervised way. Subsequently, we incorporate the retrieved samples{'} SQL into the input, enabling the model to acquire prior knowledge of similar SQL grammar. To further bridge the gap between specific and general knowledge, we present a mahalanobis contrastive learning method, which facilitates the transfer of the sample toward the specific knowledge distribution constructed by the retrieved samples. Experimental results on five datasets verify the effectiveness of our approach in improving the accuracy and robustness of Text-to-SQL generation. Our framework has achieved improved performance when combined with many other backbone models (including the 11B flan-T5) and also achieved state-of-the-art performance when compared to existing methods that employ the fine-tuning approach.",
}
| Text-to-SQL is the task that aims at translating natural language questions into SQL queries. Existing methods directly align the natural language with SQL Language and train one encoder-decoder-based model to fit all questions. However, they underestimate the inherent structural characteristics of SQL, as well as the gap between specific structure knowledge and general knowledge. This leads to structure errors in the generated SQL. To address the above challenges, we propose a retrieval-argument framework, namely ReFSQL. It contains two parts, structure-enhanced retriever and the generator. Structure-enhanced retriever is designed to identify samples with comparable specific knowledge in an unsupervised way. Subsequently, we incorporate the retrieved samples{'} SQL into the input, enabling the model to acquire prior knowledge of similar SQL grammar. To further bridge the gap between specific and general knowledge, we present a mahalanobis contrastive learning method, which facilitates the transfer of the sample toward the specific knowledge distribution constructed by the retrieved samples. Experimental results on five datasets verify the effectiveness of our approach in improving the accuracy and robustness of Text-to-SQL generation. Our framework has achieved improved performance when combined with many other backbone models (including the 11B flan-T5) and also achieved state-of-the-art performance when compared to existing methods that employ the fine-tuning approach. | [
"Zhang, Kun",
"Lin, Xiexiong",
"Wang, Yuanzhuo",
"Zhang, Xin",
"Sun, Fei",
"Jianhe, Cen",
"Tan, Hexiang",
"Jiang, Xuhui",
"Shen, Huawei"
] | ReFSQL: A Retrieval-Augmentation Framework for Text-to-SQL Generation | findings-emnlp.48 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.49.bib | https://aclanthology.org/2023.findings-emnlp.49/ | @inproceedings{csordas-etal-2023-approximating,
title = "Approximating Two-Layer Feedforward Networks for Efficient Transformers",
author = {Csord{\'a}s, R{\'o}bert and
Irie, Kazuki and
Schmidhuber, J{\"u}rgen},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.49",
doi = "10.18653/v1/2023.findings-emnlp.49",
pages = "674--692",
abstract = "How to reduce compute and memory requirements of neural networks (NNs) without sacrificing performance? Many recent works use sparse Mixtures of Experts (MoEs) to build resource-efficient large language models (LMs). Here we introduce several novel perspectives on MoEs, presenting a general framework that *unifies* various methods to *approximate two-layer NNs* (e.g., feedforward blocks of Transformers), including product-key memories (PKMs). Leveraging insights from this framework, we propose methods to improve both MoEs and PKMs. Unlike prior work that compares MoEs with dense baselines under the *compute-equal* condition, our evaluation condition is *parameter-equal*, which is crucial to properly evaluate LMs. We show that our MoEs are competitive with the *dense* Transformer-XL on both the WikiText-103 and enwiki8 datasets at two different scales, while being much more resource efficient. This demonstrates that MoEs are relevant not only to extremely large LMs but also to any-scale resource-efficient LMs. Our code is public.",
}
| How to reduce compute and memory requirements of neural networks (NNs) without sacrificing performance? Many recent works use sparse Mixtures of Experts (MoEs) to build resource-efficient large language models (LMs). Here we introduce several novel perspectives on MoEs, presenting a general framework that *unifies* various methods to *approximate two-layer NNs* (e.g., feedforward blocks of Transformers), including product-key memories (PKMs). Leveraging insights from this framework, we propose methods to improve both MoEs and PKMs. Unlike prior work that compares MoEs with dense baselines under the *compute-equal* condition, our evaluation condition is *parameter-equal*, which is crucial to properly evaluate LMs. We show that our MoEs are competitive with the *dense* Transformer-XL on both the WikiText-103 and enwiki8 datasets at two different scales, while being much more resource efficient. This demonstrates that MoEs are relevant not only to extremely large LMs but also to any-scale resource-efficient LMs. Our code is public. | [
"Csord{\\'a}s, R{\\'o}bert",
"Irie, Kazuki",
"Schmidhuber, J{\\\"u}rgen"
] | Approximating Two-Layer Feedforward Networks for Efficient Transformers | findings-emnlp.49 | 2310.10837 | [
"https://github.com/robertcsordas/moe"
] | https://huggingface.co/papers/2310.10837 | 1 | 10 | 3 | 3 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.50.bib | https://aclanthology.org/2023.findings-emnlp.50/ | @inproceedings{hu-etal-2023-adapter,
title = "Adapter-{TST}: A Parameter Efficient Method for Multiple-Attribute Text Style Transfer",
author = "Hu, Zhiqiang and
Chen, Nancy and
Lee, Roy",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.50",
doi = "10.18653/v1/2023.findings-emnlp.50",
pages = "693--703",
abstract = "Adapting a large language model for multiple-attribute text style transfer via fine-tuning can be challenging due to the substantial amount of computational resources and labeled data required for the specific downstream task. In this paper, we address this challenge by introducing Adapter-TST, a framework that freezes the pre-trained model{'}s original parameters and enables the development of a multiple-attribute text style transfer model. Using BART as the backbone model, Adapter-TST utilizes different neural adapters to model different types of attribute information, similar to a plug-in connected to BART. Our method allows control over multiple attributes (e.g. sentiment, tense, active or passive voice) and configures the adapters{'} architecture to generate multiple outputs in respect to attributes or compositional editing on the same sentence. We evaluate the proposed model on both traditional sentiment transfer and multiple-attribute transfer tasks. The experiment results demonstrate that Adapter-TST outperforms all the state-of-the-art baselines with significantly less computational resources. We have also empirically shown that each adapter is able to characterize specific stylistic attributes effectively and can be configured to perform compositional editing.",
}
| Adapting a large language model for multiple-attribute text style transfer via fine-tuning can be challenging due to the substantial amount of computational resources and labeled data required for the specific downstream task. In this paper, we address this challenge by introducing Adapter-TST, a framework that freezes the pre-trained model{'}s original parameters and enables the development of a multiple-attribute text style transfer model. Using BART as the backbone model, Adapter-TST utilizes different neural adapters to model different types of attribute information, similar to a plug-in connected to BART. Our method allows control over multiple attributes (e.g. sentiment, tense, active or passive voice) and configures the adapters{'} architecture to generate multiple outputs in respect to attributes or compositional editing on the same sentence. We evaluate the proposed model on both traditional sentiment transfer and multiple-attribute transfer tasks. The experiment results demonstrate that Adapter-TST outperforms all the state-of-the-art baselines with significantly less computational resources. We have also empirically shown that each adapter is able to characterize specific stylistic attributes effectively and can be configured to perform compositional editing. | [
"Hu, Zhiqiang",
"Chen, Nancy",
"Lee, Roy"
] | Adapter-TST: A Parameter Efficient Method for Multiple-Attribute Text Style Transfer | findings-emnlp.50 | 2305.05945 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.51.bib | https://aclanthology.org/2023.findings-emnlp.51/ | @inproceedings{gutierrez-etal-2023-solving,
title = "Solving the Right Problem is Key for Translational {NLP}: A Case Study in {UMLS} Vocabulary Insertion",
author = "Gutierrez, Bernal and
Mao, Yuqing and
Nguyen, Vinh and
Fung, Kin and
Su, Yu and
Bodenreider, Olivier",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.51",
doi = "10.18653/v1/2023.findings-emnlp.51",
pages = "704--717",
abstract = "As the immense opportunities enabled by large language models become more apparent, NLP systems will be increasingly expected to excel in real-world settings. However, in many instances, powerful models alone will not yield translational NLP solutions, especially if the formulated problem is not well aligned with the real-world task. In this work, we study the case of UMLS vocabulary insertion, an important real-world task in which hundreds of thousands of new terms, referred to as atoms, are added to the UMLS, one of the most comprehensive open-source biomedical knowledge bases. Previous work aimed to develop an automated NLP system to make this time-consuming, costly, and error-prone task more efficient. Nevertheless, practical progress in this direction has been difficult to achieve due to a problem formulation and evaluation gap between research output and the real-world task. In order to address this gap, we introduce a new formulation for UMLS vocabulary insertion which mirrors the real-world task, datasets which faithfully represent it and several strong baselines we developed through re-purposing existing solutions. Additionally, we propose an effective rule-enhanced biomedical language model which enables important new model behavior, outperforms all strong baselines and provides measurable qualitative improvements to editors who carry out the UVI task. We hope this case study provides insight into the considerable importance of problem formulation for the success of translational NLP solutions.",
}
| As the immense opportunities enabled by large language models become more apparent, NLP systems will be increasingly expected to excel in real-world settings. However, in many instances, powerful models alone will not yield translational NLP solutions, especially if the formulated problem is not well aligned with the real-world task. In this work, we study the case of UMLS vocabulary insertion, an important real-world task in which hundreds of thousands of new terms, referred to as atoms, are added to the UMLS, one of the most comprehensive open-source biomedical knowledge bases. Previous work aimed to develop an automated NLP system to make this time-consuming, costly, and error-prone task more efficient. Nevertheless, practical progress in this direction has been difficult to achieve due to a problem formulation and evaluation gap between research output and the real-world task. In order to address this gap, we introduce a new formulation for UMLS vocabulary insertion which mirrors the real-world task, datasets which faithfully represent it and several strong baselines we developed through re-purposing existing solutions. Additionally, we propose an effective rule-enhanced biomedical language model which enables important new model behavior, outperforms all strong baselines and provides measurable qualitative improvements to editors who carry out the UVI task. We hope this case study provides insight into the considerable importance of problem formulation for the success of translational NLP solutions. | [
"Gutierrez, Bernal",
"Mao, Yuqing",
"Nguyen, Vinh",
"Fung, Kin",
"Su, Yu",
"Bodenreider, Olivier"
] | Solving the Right Problem is Key for Translational NLP: A Case Study in UMLS Vocabulary Insertion | findings-emnlp.51 | 2311.15106 | [
"https://github.com/osu-nlp-group/umls-vocabulary-insertion"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.52.bib | https://aclanthology.org/2023.findings-emnlp.52/ | @inproceedings{arviv-etal-2023-improving,
title = "Improving Cross-lingual Transfer through Subtree-aware Word Reordering",
author = "Arviv, Ofir and
Nikolaev, Dmitry and
Karidi, Taelin and
Abend, Omri",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.52",
doi = "10.18653/v1/2023.findings-emnlp.52",
pages = "718--736",
abstract = "Despite the impressive growth of the abilities of multilingual language models, such as XLM-R and mT5, it has been shown that they still face difficulties when tackling typologically-distant languages, particularly in the low-resource setting. One obstacle for effective cross-lingual transfer is variability in word-order patterns. It can be potentially mitigated via source- or target-side word reordering, and numerous approaches to reordering have been proposed. However, they rely on language-specific rules, work on the level of POS tags, or only target the main clause, leaving subordinate clauses intact. To address these limitations, we present a new powerful reordering method, defined in terms of Universal Dependencies, that is able to learn fine-grained word-order patterns conditioned on the syntactic context from a small amount of annotated data and can be applied at all levels of the syntactic tree. We conduct experiments on a diverse set of tasks and show that our method consistently outperforms strong baselines over different language pairs and model architectures. This performance advantage holds true in both zero-shot and few-shot scenarios.",
}
| Despite the impressive growth of the abilities of multilingual language models, such as XLM-R and mT5, it has been shown that they still face difficulties when tackling typologically-distant languages, particularly in the low-resource setting. One obstacle for effective cross-lingual transfer is variability in word-order patterns. It can be potentially mitigated via source- or target-side word reordering, and numerous approaches to reordering have been proposed. However, they rely on language-specific rules, work on the level of POS tags, or only target the main clause, leaving subordinate clauses intact. To address these limitations, we present a new powerful reordering method, defined in terms of Universal Dependencies, that is able to learn fine-grained word-order patterns conditioned on the syntactic context from a small amount of annotated data and can be applied at all levels of the syntactic tree. We conduct experiments on a diverse set of tasks and show that our method consistently outperforms strong baselines over different language pairs and model architectures. This performance advantage holds true in both zero-shot and few-shot scenarios. | [
"Arviv, Ofir",
"Nikolaev, Dmitry",
"Karidi, Taelin",
"Abend, Omri"
] | Improving Cross-lingual Transfer through Subtree-aware Word Reordering | findings-emnlp.52 | 2310.13583 | [
"https://github.com/ofirarviv/ud-based-word-reordering"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.53.bib | https://aclanthology.org/2023.findings-emnlp.53/ | @inproceedings{liang-etal-2023-novel,
title = "Novel Slot Detection With an Incremental Setting",
author = "Liang, Chen and
Li, Hongliang and
Guan, Changhao and
Liu, Qingbin and
Liu, Jian and
Xu, Jinan and
Zhao, Zhe",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.53",
doi = "10.18653/v1/2023.findings-emnlp.53",
pages = "737--746",
abstract = "Current dialogue systems face diverse user requests and rapid change domains, making quickly adapt to scenarios with previous unseen slot types become a major challenge. Recently, researchers have introduced novel slot detection (NSD) to discover potential new types. However, dialogue system with NSD does not bring practical improvements due to the system still cannot handle novel slots in subsequent interactions. In this paper, we define incremental novel slot detection (INSD), which separates the dialogue system to deal with novel types as two major phrases: 1) model discovers unknown slots, 2) training model to possess the capability to handle new classes. We provide an effective model to extract novel slots with set prediction strategy and propose a query-enhanced approach to overcome catastrophic forgetting during the process of INSD. We construct two INSD datasets to evaluate our method and experimental results show that our approach exhibits superior performance.",
}
| Current dialogue systems face diverse user requests and rapid change domains, making quickly adapt to scenarios with previous unseen slot types become a major challenge. Recently, researchers have introduced novel slot detection (NSD) to discover potential new types. However, dialogue system with NSD does not bring practical improvements due to the system still cannot handle novel slots in subsequent interactions. In this paper, we define incremental novel slot detection (INSD), which separates the dialogue system to deal with novel types as two major phrases: 1) model discovers unknown slots, 2) training model to possess the capability to handle new classes. We provide an effective model to extract novel slots with set prediction strategy and propose a query-enhanced approach to overcome catastrophic forgetting during the process of INSD. We construct two INSD datasets to evaluate our method and experimental results show that our approach exhibits superior performance. | [
"Liang, Chen",
"Li, Hongliang",
"Guan, Changhao",
"Liu, Qingbin",
"Liu, Jian",
"Xu, Jinan",
"Zhao, Zhe"
] | Novel Slot Detection With an Incremental Setting | findings-emnlp.53 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.54.bib | https://aclanthology.org/2023.findings-emnlp.54/ | @inproceedings{jo-2023-self-supervised,
title = "Self-supervised Post-processing Method to Enrich Pretrained Word Vectors",
author = "Jo, Hwiyeol",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.54",
doi = "10.18653/v1/2023.findings-emnlp.54",
pages = "747--757",
abstract = "Retrofitting techniques, which inject external resources into word representations, have compensated for the weakness of distributed representations in semantic and relational knowledge between words. However, the previous methods require additional external resources and strongly depend on the lexicon. To address the issues, we propose a simple extension of extrofitting, self-supervised extrofitting: extrofitting by its own word vector distribution. Our methods improve the vanilla embeddings on all of word similarity tasks without any external resources. Moreover, the method is also effective in various languages, which implies that our method will be useful in lexicon-scarce languages. As downstream tasks, we show its benefits in dialogue state tracking and text classification tasks, reporting better and generalized results compared to other word vector specialization methods.",
}
| Retrofitting techniques, which inject external resources into word representations, have compensated for the weakness of distributed representations in semantic and relational knowledge between words. However, the previous methods require additional external resources and strongly depend on the lexicon. To address the issues, we propose a simple extension of extrofitting, self-supervised extrofitting: extrofitting by its own word vector distribution. Our methods improve the vanilla embeddings on all of word similarity tasks without any external resources. Moreover, the method is also effective in various languages, which implies that our method will be useful in lexicon-scarce languages. As downstream tasks, we show its benefits in dialogue state tracking and text classification tasks, reporting better and generalized results compared to other word vector specialization methods. | [
"Jo, Hwiyeol"
] | Self-supervised Post-processing Method to Enrich Pretrained Word Vectors | findings-emnlp.54 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.55.bib | https://aclanthology.org/2023.findings-emnlp.55/ | @inproceedings{zhao-etal-2023-automatic,
title = "Automatic Model Selection with Large Language Models for Reasoning",
author = "Zhao, James and
Xie, Yuxi and
Kawaguchi, Kenji and
He, Junxian and
Xie, Michael",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.55",
doi = "10.18653/v1/2023.findings-emnlp.55",
pages = "758--783",
abstract = "Chain-of-Thought (CoT) and Program-Aided Language Models (PAL) represent two distinct reasoning methods, each with its own strengths. CoT employs natural language, offering flexibility and interpretability, while PAL utilizes programming language, yielding more structured and rigorous logic. We introduce a model selection method to combine the best of both worlds by employing a large language model (LLM) to dynamically select between them. Our theoretical analysis underscores the feasibility of this method, which is further corroborated by empirical results. Our proposed method demonstrates significant performance improvements across eight reasoning datasets with Codex, ChatGPT, and GPT-4. Additionally, our method is complementary to self-consistency; when integrated, it can further enhance performance while significantly reducing computation costs. Moreover, we achieve new state-of-the-art results on GSM8K and SVAMP, with respective accuracies of 96.8{\%} and 93.7{\%}.",
}
| Chain-of-Thought (CoT) and Program-Aided Language Models (PAL) represent two distinct reasoning methods, each with its own strengths. CoT employs natural language, offering flexibility and interpretability, while PAL utilizes programming language, yielding more structured and rigorous logic. We introduce a model selection method to combine the best of both worlds by employing a large language model (LLM) to dynamically select between them. Our theoretical analysis underscores the feasibility of this method, which is further corroborated by empirical results. Our proposed method demonstrates significant performance improvements across eight reasoning datasets with Codex, ChatGPT, and GPT-4. Additionally, our method is complementary to self-consistency; when integrated, it can further enhance performance while significantly reducing computation costs. Moreover, we achieve new state-of-the-art results on GSM8K and SVAMP, with respective accuracies of 96.8{\%} and 93.7{\%}. | [
"Zhao, James",
"Xie, Yuxi",
"Kawaguchi, Kenji",
"He, Junxian",
"Xie, Michael"
] | Automatic Model Selection with Large Language Models for Reasoning | findings-emnlp.55 | 2305.14333 | [
"https://github.com/xuzhao0/model-selection-reasoning"
] | https://huggingface.co/papers/2305.14333 | 0 | 0 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.56.bib | https://aclanthology.org/2023.findings-emnlp.56/ | @inproceedings{kato-etal-2023-arkitscenerefer,
title = "{ARK}it{S}cene{R}efer: Text-based Localization of Small Objects in Diverse Real-World 3{D} Indoor Scenes",
author = "Kato, Shunya and
Kurita, Shuhei and
Chu, Chenhui and
Kurohashi, Sadao",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.56",
doi = "10.18653/v1/2023.findings-emnlp.56",
pages = "784--799",
abstract = "3D referring expression comprehension is a task to ground text representations onto objects in 3D scenes. It is a crucial task for indoor household robots or augmented reality devices to localize objects referred to in user instructions. However, existing indoor 3D referring expression comprehension datasets typically cover larger object classes that are easy to localize, such as chairs, tables, or doors, and often overlook small objects, such as cooking tools or office supplies. Based on the recently proposed diverse and high-resolution 3D scene dataset of ARKitScenes, we construct the ARKitSceneRefer dataset focusing on small daily-use objects that frequently appear in real-world indoor scenes. ARKitSceneRefer contains 15k objects of 1,605 indoor scenes, which are significantly larger than those of the existing 3D referring datasets, and covers diverse object classes of 583 from the LVIS dataset. In empirical experiments with both 2D and 3D state-of-the-art referring expression comprehension models, we observed the task difficulty of the localization in the diverse small object classes.",
}
| 3D referring expression comprehension is a task to ground text representations onto objects in 3D scenes. It is a crucial task for indoor household robots or augmented reality devices to localize objects referred to in user instructions. However, existing indoor 3D referring expression comprehension datasets typically cover larger object classes that are easy to localize, such as chairs, tables, or doors, and often overlook small objects, such as cooking tools or office supplies. Based on the recently proposed diverse and high-resolution 3D scene dataset of ARKitScenes, we construct the ARKitSceneRefer dataset focusing on small daily-use objects that frequently appear in real-world indoor scenes. ARKitSceneRefer contains 15k objects of 1,605 indoor scenes, which are significantly larger than those of the existing 3D referring datasets, and covers diverse object classes of 583 from the LVIS dataset. In empirical experiments with both 2D and 3D state-of-the-art referring expression comprehension models, we observed the task difficulty of the localization in the diverse small object classes. | [
"Kato, Shunya",
"Kurita, Shuhei",
"Chu, Chenhui",
"Kurohashi, Sadao"
] | ARKitSceneRefer: Text-based Localization of Small Objects in Diverse Real-World 3D Indoor Scenes | findings-emnlp.56 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.57.bib | https://aclanthology.org/2023.findings-emnlp.57/ | @inproceedings{xia-etal-2023-improving,
title = "Improving Question Generation with Multi-level Content Planning",
author = "Xia, Zehua and
Gou, Qi and
Yu, Bowen and
Yu, Haiyang and
Huang, Fei and
Li, Yongbin and
Cam-Tu, Nguyen",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.57",
doi = "10.18653/v1/2023.findings-emnlp.57",
pages = "800--814",
abstract = "This paper addresses the problem of generating questions from a given context and an answer, specifically focusing on questions that require multi-hop reasoning across an extended context. Previous studies have suggested that key phrase selection is essential for question generation (QG), yet it is still challenging to connect such disjointed phrases into meaningful questions, particularly for long context. To mitigate this issue, we propose MultiFactor, a novel QG framework based on multi-level content planning. Specifically, MultiFactor includes two components: FA-Model, which simultaneously selects key phrases and generates full answers, and Q-Model which takes the generated full answer as an additional input to generate questions. Here, full answer generation is introduced to connect the short answer with the selected key phrases, thus forming an answer-aware summary to facilitate QG. Both FA-Model and Q-Model are formalized as simple-yet-effective Phrase-Enhanced Transformers, our joint model for phrase selection and text generation. Experimental results show that our method outperforms strong baselines on two popular QG datasets. Our code is available at https://github.com/zeaver/MultiFactor.",
}
| This paper addresses the problem of generating questions from a given context and an answer, specifically focusing on questions that require multi-hop reasoning across an extended context. Previous studies have suggested that key phrase selection is essential for question generation (QG), yet it is still challenging to connect such disjointed phrases into meaningful questions, particularly for long context. To mitigate this issue, we propose MultiFactor, a novel QG framework based on multi-level content planning. Specifically, MultiFactor includes two components: FA-Model, which simultaneously selects key phrases and generates full answers, and Q-Model which takes the generated full answer as an additional input to generate questions. Here, full answer generation is introduced to connect the short answer with the selected key phrases, thus forming an answer-aware summary to facilitate QG. Both FA-Model and Q-Model are formalized as simple-yet-effective Phrase-Enhanced Transformers, our joint model for phrase selection and text generation. Experimental results show that our method outperforms strong baselines on two popular QG datasets. Our code is available at https://github.com/zeaver/MultiFactor. | [
"Xia, Zehua",
"Gou, Qi",
"Yu, Bowen",
"Yu, Haiyang",
"Huang, Fei",
"Li, Yongbin",
"Cam-Tu, Nguyen"
] | Improving Question Generation with Multi-level Content Planning | findings-emnlp.57 | 2310.13512 | [
"https://github.com/zeaver/multifactor"
] | https://huggingface.co/papers/2310.13512 | 1 | 0 | 0 | 7 | [] | [
"zeaver/multifactor_hotpotqa_suppfacts",
"zeaver/multifactor_squad1.1_zhou"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.58.bib | https://aclanthology.org/2023.findings-emnlp.58/ | @inproceedings{guo-etal-2023-chatgpt,
title = "Is {C}hat{GPT} a Financial Expert? Evaluating Language Models on Financial Natural Language Processing",
author = "Guo, Yue and
Xu, Zian and
Yang, Yi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.58",
doi = "10.18653/v1/2023.findings-emnlp.58",
pages = "815--821",
abstract = "The emergence of Large Language Models (LLMs), such as ChatGPT, has revolutionized general natural language preprocessing (NLP) tasks. However, their expertise in the financial domain lacks a comprehensive evaluation. To assess the ability of LLMs to solve financial NLP tasks, we present FinLMEval, a framework for Financial Language Model Evaluation, comprising nine datasets designed to evaluate the performance of language models. This study compares the performance of fine-tuned auto-encoding language models (BERT, RoBERTa, FinBERT) and the LLM ChatGPT. Our findings reveal that while ChatGPT demonstrates notable performance across most financial tasks, it generally lags behind the fine-tuned expert models, especially when dealing with proprietary datasets. We hope this study builds foundation evaluation benchmarks for continuing efforts to build more advanced LLMs in the financial domain.",
}
| The emergence of Large Language Models (LLMs), such as ChatGPT, has revolutionized general natural language preprocessing (NLP) tasks. However, their expertise in the financial domain lacks a comprehensive evaluation. To assess the ability of LLMs to solve financial NLP tasks, we present FinLMEval, a framework for Financial Language Model Evaluation, comprising nine datasets designed to evaluate the performance of language models. This study compares the performance of fine-tuned auto-encoding language models (BERT, RoBERTa, FinBERT) and the LLM ChatGPT. Our findings reveal that while ChatGPT demonstrates notable performance across most financial tasks, it generally lags behind the fine-tuned expert models, especially when dealing with proprietary datasets. We hope this study builds foundation evaluation benchmarks for continuing efforts to build more advanced LLMs in the financial domain. | [
"Guo, Yue",
"Xu, Zian",
"Yang, Yi"
] | Is ChatGPT a Financial Expert? Evaluating Language Models on Financial Natural Language Processing | findings-emnlp.58 | 2310.12664 | [
""
] | https://huggingface.co/papers/2310.12664 | 0 | 0 | 0 | 3 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.59.bib | https://aclanthology.org/2023.findings-emnlp.59/ | @inproceedings{sadat-etal-2023-delucionqa,
title = "{D}elucion{QA}: Detecting Hallucinations in Domain-specific Question Answering",
author = "Sadat, Mobashir and
Zhou, Zhengyu and
Lange, Lukas and
Araki, Jun and
Gundroo, Arsalan and
Wang, Bingqing and
Menon, Rakesh and
Parvez, Md and
Feng, Zhe",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.59",
doi = "10.18653/v1/2023.findings-emnlp.59",
pages = "822--835",
abstract = "Hallucination is a well-known phenomenon in text generated by large language models (LLMs). The existence of hallucinatory responses is found in almost all application scenarios e.g., summarization, question-answering (QA) etc. For applications requiring high reliability (e.g., customer-facing assistants), the potential existence of hallucination in LLM-generated text is a critical problem. The amount of hallucination can be reduced by leveraging information retrieval to provide relevant background information to the LLM. However, LLMs can still generate hallucinatory content for various reasons (e.g., prioritizing its parametric knowledge over the context, failure to capture the relevant information from the context, etc.). Detecting hallucinations through automated methods is thus paramount. To facilitate research in this direction, we introduce a sophisticated dataset, DelucionQA, that captures hallucinations made by retrieval-augmented LLMs for a domain-specific QA task. Furthermore, we propose a set of hallucination detection methods to serve as baselines for future works from the research community. Analysis and case study are also provided to share valuable insights on hallucination phenomena in the target scenario.",
}
| Hallucination is a well-known phenomenon in text generated by large language models (LLMs). The existence of hallucinatory responses is found in almost all application scenarios e.g., summarization, question-answering (QA) etc. For applications requiring high reliability (e.g., customer-facing assistants), the potential existence of hallucination in LLM-generated text is a critical problem. The amount of hallucination can be reduced by leveraging information retrieval to provide relevant background information to the LLM. However, LLMs can still generate hallucinatory content for various reasons (e.g., prioritizing its parametric knowledge over the context, failure to capture the relevant information from the context, etc.). Detecting hallucinations through automated methods is thus paramount. To facilitate research in this direction, we introduce a sophisticated dataset, DelucionQA, that captures hallucinations made by retrieval-augmented LLMs for a domain-specific QA task. Furthermore, we propose a set of hallucination detection methods to serve as baselines for future works from the research community. Analysis and case study are also provided to share valuable insights on hallucination phenomena in the target scenario. | [
"Sadat, Mobashir",
"Zhou, Zhengyu",
"Lange, Lukas",
"Araki, Jun",
"Gundroo, Arsalan",
"Wang, Bingqing",
"Menon, Rakesh",
"Parvez, Md",
"Feng, Zhe"
] | DelucionQA: Detecting Hallucinations in Domain-specific Question Answering | findings-emnlp.59 | 2312.05200 | [
""
] | https://huggingface.co/papers/2312.05200 | 1 | 1 | 0 | 9 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.60.bib | https://aclanthology.org/2023.findings-emnlp.60/ | @inproceedings{jian-wang-2023-invgc,
title = "{I}nv{GC}: Robust Cross-Modal Retrieval by Inverse Graph Convolution",
author = "Jian, Xiangru and
Wang, Yimu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.60",
doi = "10.18653/v1/2023.findings-emnlp.60",
pages = "836--865",
abstract = "Over recent decades, significant advancements in cross-modal retrieval is mainly driven by breakthroughs in visual and linguistic modeling. However, a recent study shows that multi-modal data representations tend to cluster within a limited convex cone (as representation degeneration problem), which hinders retrieval performance due to the inseparability of these representations. In our study, we first empirically validate the presence of the representation degeneration problem across multiple cross-modal benchmarks and methods. Next, to address it, we introduce a novel method, called InvGC, a post-processing technique inspired by graph convolution and average pooling. Specifically, InvGC defines the graph topology within the datasets and then applies graph convolution in a subtractive manner. This method effectively separates representations by increasing the distances between data points. To improve the efficiency and effectiveness of InvGC, we propose an advanced graph topology, LocalAdj, which only aims to increase the distances between each data point and its nearest neighbors. To understand why InvGC works, we present a detailed theoretical analysis, proving that the lower bound of recall will be improved after deploying InvGC. Extensive empirical results show that InvGC and InvGC w/LocalAdj significantly mitigate the representation degeneration problem, thereby enhancing retrieval performance.",
}
| Over recent decades, significant advancements in cross-modal retrieval is mainly driven by breakthroughs in visual and linguistic modeling. However, a recent study shows that multi-modal data representations tend to cluster within a limited convex cone (as representation degeneration problem), which hinders retrieval performance due to the inseparability of these representations. In our study, we first empirically validate the presence of the representation degeneration problem across multiple cross-modal benchmarks and methods. Next, to address it, we introduce a novel method, called InvGC, a post-processing technique inspired by graph convolution and average pooling. Specifically, InvGC defines the graph topology within the datasets and then applies graph convolution in a subtractive manner. This method effectively separates representations by increasing the distances between data points. To improve the efficiency and effectiveness of InvGC, we propose an advanced graph topology, LocalAdj, which only aims to increase the distances between each data point and its nearest neighbors. To understand why InvGC works, we present a detailed theoretical analysis, proving that the lower bound of recall will be improved after deploying InvGC. Extensive empirical results show that InvGC and InvGC w/LocalAdj significantly mitigate the representation degeneration problem, thereby enhancing retrieval performance. | [
"Jian, Xiangru",
"Wang, Yimu"
] | InvGC: Robust Cross-Modal Retrieval by Inverse Graph Convolution | findings-emnlp.60 | 2310.13276 | [
"https://github.com/yimuwangcs/Better_Cross_Modal_Retrieval"
] | https://huggingface.co/papers/2310.13276 | 1 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.61.bib | https://aclanthology.org/2023.findings-emnlp.61/ | @inproceedings{raunak-etal-2023-dissecting,
title = "Dissecting In-Context Learning of Translations in {GPT}-3",
author = "Raunak, Vikas and
Menezes, Arul and
Awadalla, Hany",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.61",
doi = "10.18653/v1/2023.findings-emnlp.61",
pages = "866--872",
abstract = "Most of the recent work in leveraging Large Language Models (LLMs) such as GPT-3 for Machine Translation (MT) has focused on selecting the few-shot samples for prompting. In this work, we try to better understand the role of demonstration attributes for the in-context learning of translations through perturbations of high-quality, in-domain demonstrations. We find that asymmetric perturbation of the source-target mappings yield vastly different results. We show that the perturbation of the source side has surprisingly little impact, while target perturbation can drastically reduce translation quality, suggesting that it is the output text distribution that provides the most important learning signal during in-context learning of translations. We propose a method named Zero-Shot-Context to add this signal automatically in Zero-Shot prompting. We demonstrate that it improves upon the zero-shot translation performance of GPT-3, even making it competitive with few-shot prompted translations.",
}
| Most of the recent work in leveraging Large Language Models (LLMs) such as GPT-3 for Machine Translation (MT) has focused on selecting the few-shot samples for prompting. In this work, we try to better understand the role of demonstration attributes for the in-context learning of translations through perturbations of high-quality, in-domain demonstrations. We find that asymmetric perturbation of the source-target mappings yield vastly different results. We show that the perturbation of the source side has surprisingly little impact, while target perturbation can drastically reduce translation quality, suggesting that it is the output text distribution that provides the most important learning signal during in-context learning of translations. We propose a method named Zero-Shot-Context to add this signal automatically in Zero-Shot prompting. We demonstrate that it improves upon the zero-shot translation performance of GPT-3, even making it competitive with few-shot prompted translations. | [
"Raunak, Vikas",
"Menezes, Arul",
"Awadalla, Hany"
] | Dissecting In-Context Learning of Translations in GPT-3 | findings-emnlp.61 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.62.bib | https://aclanthology.org/2023.findings-emnlp.62/ | @inproceedings{reddy-etal-2023-social,
title = "Social Commonsense-Guided Search Query Generation for Open-Domain Knowledge-Powered Conversations",
author = "Reddy, Revanth and
Bai, Hao and
Yao, Wentao and
Suresh, Sharath Chandra Etagi and
Ji, Heng and
Zhai, ChengXiang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.62",
doi = "10.18653/v1/2023.findings-emnlp.62",
pages = "873--885",
abstract = "Open-domain dialog involves generating search queries that help obtain relevant knowledge for holding informative conversations. However, it can be challenging to determine what information to retrieve when the user is passive and does not express a clear need or request. To tackle this issue, we present a novel approach that focuses on generating internet search queries that are guided by social commonsense. Specifically, we leverage a commonsense dialog system to establish connections related to the conversation topic, which subsequently guides our query generation. Our proposed framework addresses passive user interactions by integrating topic tracking, commonsense response generation and instruction-driven query generation. Through extensive evaluations, we show that our approach overcomes limitations of existing query generation techniques that rely solely on explicit dialog information, and produces search queries that are more relevant, specific, and compelling, ultimately resulting in more engaging responses.",
}
| Open-domain dialog involves generating search queries that help obtain relevant knowledge for holding informative conversations. However, it can be challenging to determine what information to retrieve when the user is passive and does not express a clear need or request. To tackle this issue, we present a novel approach that focuses on generating internet search queries that are guided by social commonsense. Specifically, we leverage a commonsense dialog system to establish connections related to the conversation topic, which subsequently guides our query generation. Our proposed framework addresses passive user interactions by integrating topic tracking, commonsense response generation and instruction-driven query generation. Through extensive evaluations, we show that our approach overcomes limitations of existing query generation techniques that rely solely on explicit dialog information, and produces search queries that are more relevant, specific, and compelling, ultimately resulting in more engaging responses. | [
"Reddy, Revanth",
"Bai, Hao",
"Yao, Wentao",
"Suresh, Sharath Ch",
"ra Etagi",
"Ji, Heng",
"Zhai, ChengXiang"
] | Social Commonsense-Guided Search Query Generation for Open-Domain Knowledge-Powered Conversations | findings-emnlp.62 | 2310.14340 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.63.bib | https://aclanthology.org/2023.findings-emnlp.63/ | @inproceedings{xie-etal-2023-mixtea,
title = "{M}ix{TEA}: Semi-supervised Entity Alignment with Mixture Teaching",
author = "Xie, Feng and
Song, Xin and
Zeng, Xiang and
Zhao, Xuechen and
Tian, Lei and
Zhou, Bin and
Tan, Yusong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.63",
doi = "10.18653/v1/2023.findings-emnlp.63",
pages = "886--896",
abstract = "Semi-supervised entity alignment (EA) is a practical and challenging task because of the lack of adequate labeled mappings as training data. Most works address this problem by generating pseudo mappings for unlabeled entities. However, they either suffer from the erroneous (noisy) pseudo mappings or largely ignore the uncertainty of pseudo mappings. In this paper, we propose a novel semi-supervised EA method, termed as MixTEA, which guides the model learning with an end-to-end mixture teaching of manually labeled mappings and probabilistic pseudo mappings. We firstly train a student model using few labeled mappings as standard. More importantly, in pseudo mapping learning, we propose a bi-directional voting (BDV) strategy that fuses the alignment decisions in different directions to estimate the uncertainty via the joint matching confidence score. Meanwhile, we also design a matching diversity-based rectification (MDR) module to adjust the pseudo mapping learning, thus reducing the negative influence of noisy mappings. Extensive results on benchmark datasets as well as further analyses demonstrate the superiority and the effectiveness of our proposed method.",
}
| Semi-supervised entity alignment (EA) is a practical and challenging task because of the lack of adequate labeled mappings as training data. Most works address this problem by generating pseudo mappings for unlabeled entities. However, they either suffer from the erroneous (noisy) pseudo mappings or largely ignore the uncertainty of pseudo mappings. In this paper, we propose a novel semi-supervised EA method, termed as MixTEA, which guides the model learning with an end-to-end mixture teaching of manually labeled mappings and probabilistic pseudo mappings. We firstly train a student model using few labeled mappings as standard. More importantly, in pseudo mapping learning, we propose a bi-directional voting (BDV) strategy that fuses the alignment decisions in different directions to estimate the uncertainty via the joint matching confidence score. Meanwhile, we also design a matching diversity-based rectification (MDR) module to adjust the pseudo mapping learning, thus reducing the negative influence of noisy mappings. Extensive results on benchmark datasets as well as further analyses demonstrate the superiority and the effectiveness of our proposed method. | [
"Xie, Feng",
"Song, Xin",
"Zeng, Xiang",
"Zhao, Xuechen",
"Tian, Lei",
"Zhou, Bin",
"Tan, Yusong"
] | MixTEA: Semi-supervised Entity Alignment with Mixture Teaching | findings-emnlp.63 | 2311.04441 | [
"https://github.com/xiefeng69/mixtea"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.65.bib | https://aclanthology.org/2023.findings-emnlp.65/ | @inproceedings{jiang-etal-2023-boot,
title = "Boot and Switch: Alternating Distillation for Zero-Shot Dense Retrieval",
author = "Jiang, Fan and
Xu, Qiongkai and
Drummond, Tom and
Cohn, Trevor",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.65",
doi = "10.18653/v1/2023.findings-emnlp.65",
pages = "912--931",
abstract = "Neural {`}dense{'} retrieval models are state of the art for many datasets, however these models often exhibit limited domain transfer ability. Existing approaches to adaptation are unwieldy, such as requiring explicit supervision, complex model architectures, or massive external models. We present $\texttt{ABEL}$, a simple but effective unsupervised method to enhance passage retrieval in zero-shot settings. Our technique follows a straightforward loop: a dense retriever learns from supervision signals provided by a reranker, and subsequently, the reranker is updated based on feedback from the improved retriever. By iterating this loop, the two components mutually enhance one another{'}s performance. Experimental results demonstrate that our unsupervised $\texttt{ABEL}$ model outperforms both leading supervised and unsupervised retrievers on the BEIR benchmark. Meanwhile, it exhibits strong adaptation abilities to tasks and domains that were unseen during training. By either fine-tuning $\texttt{ABEL}$ on labelled data or integrating it with existing supervised dense retrievers, we achieve state-of-the-art results.",
}
| Neural {`}dense{'} retrieval models are state of the art for many datasets, however these models often exhibit limited domain transfer ability. Existing approaches to adaptation are unwieldy, such as requiring explicit supervision, complex model architectures, or massive external models. We present $\texttt{ABEL}$, a simple but effective unsupervised method to enhance passage retrieval in zero-shot settings. Our technique follows a straightforward loop: a dense retriever learns from supervision signals provided by a reranker, and subsequently, the reranker is updated based on feedback from the improved retriever. By iterating this loop, the two components mutually enhance one another{'}s performance. Experimental results demonstrate that our unsupervised $\texttt{ABEL}$ model outperforms both leading supervised and unsupervised retrievers on the BEIR benchmark. Meanwhile, it exhibits strong adaptation abilities to tasks and domains that were unseen during training. By either fine-tuning $\texttt{ABEL}$ on labelled data or integrating it with existing supervised dense retrievers, we achieve state-of-the-art results. | [
"Jiang, Fan",
"Xu, Qiongkai",
"Drummond, Tom",
"Cohn, Trevor"
] | Boot and Switch: Alternating Distillation for Zero-Shot Dense Retrieval | findings-emnlp.65 | 2311.15564 | [
"https://github.com/fantabulous-j/bootswitch"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.66.bib | https://aclanthology.org/2023.findings-emnlp.66/ | @inproceedings{ren-etal-2023-testa,
title = "{TESTA}: Temporal-Spatial Token Aggregation for Long-form Video-Language Understanding",
author = "Ren, Shuhuai and
Chen, Sishuo and
Li, Shicheng and
Sun, Xu and
Hou, Lu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.66",
doi = "10.18653/v1/2023.findings-emnlp.66",
pages = "932--947",
abstract = "Large-scale video-language pre-training has made remarkable strides in advancing video-language understanding tasks. However, the heavy computational burden of video encoding remains a formidable efficiency bottleneck, particularly for long-form videos. These videos contain massive visual tokens due to their inherent 3D properties and spatiotemporal redundancy, making it challenging to capture complex temporal and spatial relationships. To tackle this issue, we propose an efficient method called TEmporal-Spatial Token Aggregation (TESTA). TESTA condenses video semantics by adaptively aggregating similar frames, as well as similar patches within each frame. TESTA can reduce the number of visual tokens by 75{\%} and thus accelerate video encoding. Building upon TESTA, we introduce a pre-trained video-language model equipped with a divided space-time token aggregation module in each video encoder block. We evaluate our model on five datasets for paragraph-to-video retrieval and long-form VideoQA tasks. Experimental results show that TESTA improves computing efficiency by 1.7 times, and achieves significant performance gains from its scalability in processing longer input frames, e.g., +13.7 R@1 on QuerYD and +6.5 R@1 on Condensed Movie.",
}
| Large-scale video-language pre-training has made remarkable strides in advancing video-language understanding tasks. However, the heavy computational burden of video encoding remains a formidable efficiency bottleneck, particularly for long-form videos. These videos contain massive visual tokens due to their inherent 3D properties and spatiotemporal redundancy, making it challenging to capture complex temporal and spatial relationships. To tackle this issue, we propose an efficient method called TEmporal-Spatial Token Aggregation (TESTA). TESTA condenses video semantics by adaptively aggregating similar frames, as well as similar patches within each frame. TESTA can reduce the number of visual tokens by 75{\%} and thus accelerate video encoding. Building upon TESTA, we introduce a pre-trained video-language model equipped with a divided space-time token aggregation module in each video encoder block. We evaluate our model on five datasets for paragraph-to-video retrieval and long-form VideoQA tasks. Experimental results show that TESTA improves computing efficiency by 1.7 times, and achieves significant performance gains from its scalability in processing longer input frames, e.g., +13.7 R@1 on QuerYD and +6.5 R@1 on Condensed Movie. | [
"Ren, Shuhuai",
"Chen, Sishuo",
"Li, Shicheng",
"Sun, Xu",
"Hou, Lu"
] | TESTA: Temporal-Spatial Token Aggregation for Long-form Video-Language Understanding | findings-emnlp.66 | 2310.19060 | [
"https://github.com/renshuhuai-andy/testa"
] | https://huggingface.co/papers/2310.19060 | 1 | 0 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.67.bib | https://aclanthology.org/2023.findings-emnlp.67/ | @inproceedings{su-etal-2023-fusing,
title = "Fusing Temporal Graphs into Transformers for Time-Sensitive Question Answering",
author = "Su, Xin and
Howard, Phillip and
Hakim, Nagib and
Bethard, Steven",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.67",
doi = "10.18653/v1/2023.findings-emnlp.67",
pages = "948--966",
abstract = "Answering time-sensitive questions from long documents requires temporal reasoning over the times in questions and documents. An important open question is whether large language models can perform such reasoning solely using a provided text document, or whether they can benefit from additional temporal information extracted using other systems. We address this research question by applying existing temporal information extraction systems to construct temporal graphs of events, times, and temporal relations in questions and documents. We then investigate different approaches for fusing these graphs into Transformer models. Experimental results show that our proposed approach for fusing temporal graphs into input text substantially enhances the temporal reasoning capabilities of Transformer models with or without fine-tuning. Additionally, our proposed method outperforms various graph convolution-based approaches and establishes a new state-of-the-art performance on SituatedQA and three splits of TimeQA.",
}
| Answering time-sensitive questions from long documents requires temporal reasoning over the times in questions and documents. An important open question is whether large language models can perform such reasoning solely using a provided text document, or whether they can benefit from additional temporal information extracted using other systems. We address this research question by applying existing temporal information extraction systems to construct temporal graphs of events, times, and temporal relations in questions and documents. We then investigate different approaches for fusing these graphs into Transformer models. Experimental results show that our proposed approach for fusing temporal graphs into input text substantially enhances the temporal reasoning capabilities of Transformer models with or without fine-tuning. Additionally, our proposed method outperforms various graph convolution-based approaches and establishes a new state-of-the-art performance on SituatedQA and three splits of TimeQA. | [
"Su, Xin",
"Howard, Phillip",
"Hakim, Nagib",
"Bethard, Steven"
] | Fusing Temporal Graphs into Transformers for Time-Sensitive Question Answering | findings-emnlp.67 | 2310.19292 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.68.bib | https://aclanthology.org/2023.findings-emnlp.68/ | @inproceedings{azaria-mitchell-2023-internal,
title = "The Internal State of an {LLM} Knows When It{'}s Lying",
author = "Azaria, Amos and
Mitchell, Tom",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.68",
doi = "10.18653/v1/2023.findings-emnlp.68",
pages = "967--976",
abstract = "While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM{'}s internal state can be used to reveal the truthfulness of statements. This includes both statements provided to the LLM, and statements that the LLM itself generates. Our approach is to train a classifier that outputs the probability that a statement is truthful, based on the hidden layer activations of the LLM as it reads or generates the statement. Experiments demonstrate that given a set of test sentences, of which half are true and half false, our trained classifier achieves an average of 71{\%} to 83{\%} accuracy labeling which sentences are true versus false, depending on the LLM base model. Furthermore, we explore the relationship between our classifier{'}s performance and approaches based on the probability assigned to the sentence by the LLM. We show that while LLM-assigned sentence probability is related to sentence truthfulness, this probability is also dependent on sentence length and the frequencies of words in the sentence, resulting in our trained classifier providing a more reliable approach to detecting truthfulness, highlighting its potential to enhance the reliability of LLM-generated content and its practical applicability in real-world scenarios.",
}
| While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM{'}s internal state can be used to reveal the truthfulness of statements. This includes both statements provided to the LLM, and statements that the LLM itself generates. Our approach is to train a classifier that outputs the probability that a statement is truthful, based on the hidden layer activations of the LLM as it reads or generates the statement. Experiments demonstrate that given a set of test sentences, of which half are true and half false, our trained classifier achieves an average of 71{\%} to 83{\%} accuracy labeling which sentences are true versus false, depending on the LLM base model. Furthermore, we explore the relationship between our classifier{'}s performance and approaches based on the probability assigned to the sentence by the LLM. We show that while LLM-assigned sentence probability is related to sentence truthfulness, this probability is also dependent on sentence length and the frequencies of words in the sentence, resulting in our trained classifier providing a more reliable approach to detecting truthfulness, highlighting its potential to enhance the reliability of LLM-generated content and its practical applicability in real-world scenarios. | [
"Azaria, Amos",
"Mitchell, Tom"
] | The Internal State of an LLM Knows When It's Lying | findings-emnlp.68 | 2304.13734 | [
""
] | https://huggingface.co/papers/2304.13734 | 0 | 2 | 1 | 2 | [
"leondz/refutation_detector_distilbert"
] | [] | [
"Tumeryk-Inc/model-security"
] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.69.bib | https://aclanthology.org/2023.findings-emnlp.69/ | @inproceedings{gao-etal-2023-factual,
title = "Factual Relation Discrimination for Factuality-oriented Abstractive Summarization",
author = "Gao, Zhiguang and
Li, Peifeng and
Jiang, Feng and
Chu, Xiaomin and
Zhu, Qiaoming",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.69",
doi = "10.18653/v1/2023.findings-emnlp.69",
pages = "977--986",
abstract = "Most neural abstractive summarization models are capable of producing high-quality summaries. However, they still frequently contain factual errors. Existing factuality-oriented abstractive summarization models only consider the integration of factual information and ignore the causes of factual errors. To address this issue, we propose a factuality-oriented abstractive summarization model DASum, which is based on a new task factual relation discrimination that is able to identify the causes of factual errors. First, we use data augmentation methods to construct counterfactual summaries (i. e., negative samples), and build a factual summarization dataset. Then, we propose the factual relation discrimination task, which determines the factuality of the dependency relations in summaries during summary generation and guides our DASum to generate factual relations, thereby improving the factuality of summaries. Experimental results on the CNN/DM and XSUM datasets show that our DASum outperforms several state-of-the-art benchmarks in terms of the factual metrics.",
}
| Most neural abstractive summarization models are capable of producing high-quality summaries. However, they still frequently contain factual errors. Existing factuality-oriented abstractive summarization models only consider the integration of factual information and ignore the causes of factual errors. To address this issue, we propose a factuality-oriented abstractive summarization model DASum, which is based on a new task factual relation discrimination that is able to identify the causes of factual errors. First, we use data augmentation methods to construct counterfactual summaries (i. e., negative samples), and build a factual summarization dataset. Then, we propose the factual relation discrimination task, which determines the factuality of the dependency relations in summaries during summary generation and guides our DASum to generate factual relations, thereby improving the factuality of summaries. Experimental results on the CNN/DM and XSUM datasets show that our DASum outperforms several state-of-the-art benchmarks in terms of the factual metrics. | [
"Gao, Zhiguang",
"Li, Peifeng",
"Jiang, Feng",
"Chu, Xiaomin",
"Zhu, Qiaoming"
] | Factual Relation Discrimination for Factuality-oriented Abstractive Summarization | findings-emnlp.69 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.70.bib | https://aclanthology.org/2023.findings-emnlp.70/ | @inproceedings{li-etal-2023-multi-modal-knowledge,
title = "Multi-Modal Knowledge Graph Transformer Framework for Multi-Modal Entity Alignment",
author = "Li, Qian and
Ji, Cheng and
Guo, Shu and
Liang, Zhaoji and
Wang, Lihong and
Li, Jianxin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.70",
doi = "10.18653/v1/2023.findings-emnlp.70",
pages = "987--999",
abstract = "Multi-Modal Entity Alignment (MMEA) is a critical task that aims to identify equivalent entity pairs across multi-modal knowledge graphs (MMKGs). However, this task faces challenges due to the presence of different types of information, including neighboring entities, multi-modal attributes, and entity types. Directly incorporating the above information (e.g., concatenation or attention) can lead to an unaligned information space. To address these challenges, we propose a novel MMEA transformer, called Meaformer, that hierarchically introduces neighbor features, multi-modal attributes, and entity types to enhance the alignment task. Taking advantage of the transformer{'}s ability to better integrate multiple information, we design a hierarchical modifiable self-attention block in a transformer encoder to preserve the unique semantics of different information. Furthermore, we design two entity-type prefix injection methods to redintegrate entity-type information using type prefixes, which help to restrict the global information of entities not present in the MMKGs.",
}
| Multi-Modal Entity Alignment (MMEA) is a critical task that aims to identify equivalent entity pairs across multi-modal knowledge graphs (MMKGs). However, this task faces challenges due to the presence of different types of information, including neighboring entities, multi-modal attributes, and entity types. Directly incorporating the above information (e.g., concatenation or attention) can lead to an unaligned information space. To address these challenges, we propose a novel MMEA transformer, called Meaformer, that hierarchically introduces neighbor features, multi-modal attributes, and entity types to enhance the alignment task. Taking advantage of the transformer{'}s ability to better integrate multiple information, we design a hierarchical modifiable self-attention block in a transformer encoder to preserve the unique semantics of different information. Furthermore, we design two entity-type prefix injection methods to redintegrate entity-type information using type prefixes, which help to restrict the global information of entities not present in the MMKGs. | [
"Li, Qian",
"Ji, Cheng",
"Guo, Shu",
"Liang, Zhaoji",
"Wang, Lihong",
"Li, Jianxin"
] | Multi-Modal Knowledge Graph Transformer Framework for Multi-Modal Entity Alignment | findings-emnlp.70 | 2310.06365 | [
"https://github.com/xiaoqian19940510/moalign"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.71.bib | https://aclanthology.org/2023.findings-emnlp.71/ | @inproceedings{libovicky-2023-prestigious,
title = "Is a Prestigious Job the same as a Prestigious Country? A Case Study on Multilingual Sentence Embeddings and {E}uropean Countries",
author = "Libovick{\'y}, Jind{\v{r}}ich",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.71",
doi = "10.18653/v1/2023.findings-emnlp.71",
pages = "1000--1010",
abstract = "We study how multilingual sentence representations capture European countries and occupations and how this differs across European languages. We prompt the models with templated sentences that we machine-translate into 12 European languages and analyze the most prominent dimensions in the embeddings. Our analysis reveals that the most prominent feature in the embedding is the political distinction between Eastern and Western Europe and the country{'}s economic strength in terms of GDP. When prompted specifically for job prestige, the embedding space clearly distinguishes high and low-prestige jobs. The occupational dimension is uncorrelated with the most dominant country dimensions in three out of four studied models. The exception is a small distilled model that exhibits a connection between occupational prestige and country of origin, which is a potential source of nationality-based discrimination. Our findings are consistent across languages.",
}
| We study how multilingual sentence representations capture European countries and occupations and how this differs across European languages. We prompt the models with templated sentences that we machine-translate into 12 European languages and analyze the most prominent dimensions in the embeddings. Our analysis reveals that the most prominent feature in the embedding is the political distinction between Eastern and Western Europe and the country{'}s economic strength in terms of GDP. When prompted specifically for job prestige, the embedding space clearly distinguishes high and low-prestige jobs. The occupational dimension is uncorrelated with the most dominant country dimensions in three out of four studied models. The exception is a small distilled model that exhibits a connection between occupational prestige and country of origin, which is a potential source of nationality-based discrimination. Our findings are consistent across languages. | [
"Libovick{\\'y}, Jind{\\v{r}}ich"
] | Is a Prestigious Job the same as a Prestigious Country? A Case Study on Multilingual Sentence Embeddings and European Countries | findings-emnlp.71 | 2305.14482 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.72.bib | https://aclanthology.org/2023.findings-emnlp.72/ | @inproceedings{ma-etal-2023-towards-holistic,
title = "Towards A Holistic Landscape of Situated Theory of Mind in Large Language Models",
author = "Ma, Ziqiao and
Sansom, Jacob and
Peng, Run and
Chai, Joyce",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.72",
doi = "10.18653/v1/2023.findings-emnlp.72",
pages = "1011--1031",
abstract = "Large Language Models (LLMs) have generated considerable interest and debate regarding their potential emergence of Theory of Mind (ToM). Several recent inquiries reveal a lack of robust ToM in these models and pose a pressing demand to develop new benchmarks, as current ones primarily focus on different aspects of ToM and are prone to shortcuts and data leakage. In this position paper, we seek to answer two road-blocking questions: (1) How can we taxonomize a holistic landscape of machine ToM? (2) What is a more effective evaluation protocol for machine ToM? Following psychological studies, we taxonomize machine ToM into 7 mental state categories and delineate existing benchmarks to identify under-explored aspects of ToM. We argue for a holistic and situated evaluation of ToM to break ToM into individual components and treat LLMs as an agent who is physically situated in environments and socially situated in interactions with humans. Such situated evaluation provides a more comprehensive assessment of mental states and potentially mitigates the risk of shortcuts and data leakage. We further present a pilot study in a grid world setup as a proof of concept. We hope this position paper can facilitate future research to integrate ToM with LLMs and offer an intuitive means for researchers to better position their work in the landscape of ToM.",
}
| Large Language Models (LLMs) have generated considerable interest and debate regarding their potential emergence of Theory of Mind (ToM). Several recent inquiries reveal a lack of robust ToM in these models and pose a pressing demand to develop new benchmarks, as current ones primarily focus on different aspects of ToM and are prone to shortcuts and data leakage. In this position paper, we seek to answer two road-blocking questions: (1) How can we taxonomize a holistic landscape of machine ToM? (2) What is a more effective evaluation protocol for machine ToM? Following psychological studies, we taxonomize machine ToM into 7 mental state categories and delineate existing benchmarks to identify under-explored aspects of ToM. We argue for a holistic and situated evaluation of ToM to break ToM into individual components and treat LLMs as an agent who is physically situated in environments and socially situated in interactions with humans. Such situated evaluation provides a more comprehensive assessment of mental states and potentially mitigates the risk of shortcuts and data leakage. We further present a pilot study in a grid world setup as a proof of concept. We hope this position paper can facilitate future research to integrate ToM with LLMs and offer an intuitive means for researchers to better position their work in the landscape of ToM. | [
"Ma, Ziqiao",
"Sansom, Jacob",
"Peng, Run",
"Chai, Joyce"
] | Towards A Holistic Landscape of Situated Theory of Mind in Large Language Models | findings-emnlp.72 | 2310.19619 | [
"https://github.com/mars-tin/awesome-theory-of-mind"
] | https://huggingface.co/papers/2310.19619 | 1 | 0 | 0 | 4 | [] | [
"sled-umich/2D-ATOMS"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.73.bib | https://aclanthology.org/2023.findings-emnlp.73/ | @inproceedings{suo-etal-2023-text,
title = "Text Augmented Spatial Aware Zero-shot Referring Image Segmentation",
author = "Suo, Yucheng and
Zhu, Linchao and
Yang, Yi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.73",
doi = "10.18653/v1/2023.findings-emnlp.73",
pages = "1032--1043",
abstract = "In this paper, we study a challenging task of zero-shot referring image segmentation. This task aims to identify the instance mask that is most related to a referring expression \textbf{without} training on pixel-level annotations. Previous research takes advantage of pre-trained cross-modal models, e.g., CLIP, to align instance-level masks with referring expressions. Yet, CLIP only considers the global-level alignment of image-text pairs, neglecting fine-grained matching between the referring sentence and local image regions. To address this challenge, we introduce a Text Augmented Spatial-aware (TAS) zero-shot referring image segmentation framework that is training-free and robust to various visual encoders. TAS incorporates a mask proposal network for instance-level mask extraction, a text-augmented visual-text matching score for mining the image-text correlation, and a spatial rectifier for mask post-processing. Notably, the text-augmented visual-text matching score leverages a $P$-score and an $N$-score in addition to the typical visual-text matching score. The $P$-score is utilized to close the visual-text domain gap through a surrogate captioning model, where the score is computed between the surrogate model-generated texts and the referring expression. The $N$-score considers the fine-grained alignment of region-text pairs via negative phrase mining, encouraging the masked image to be repelled from the mined distracting phrases. Extensive experiments are conducted on various datasets, including RefCOCO, RefCOCO+, and RefCOCOg. The proposed method clearly outperforms state-of-the-art zero-shot referring image segmentation methods.",
}
| In this paper, we study a challenging task of zero-shot referring image segmentation. This task aims to identify the instance mask that is most related to a referring expression \textbf{without} training on pixel-level annotations. Previous research takes advantage of pre-trained cross-modal models, e.g., CLIP, to align instance-level masks with referring expressions. Yet, CLIP only considers the global-level alignment of image-text pairs, neglecting fine-grained matching between the referring sentence and local image regions. To address this challenge, we introduce a Text Augmented Spatial-aware (TAS) zero-shot referring image segmentation framework that is training-free and robust to various visual encoders. TAS incorporates a mask proposal network for instance-level mask extraction, a text-augmented visual-text matching score for mining the image-text correlation, and a spatial rectifier for mask post-processing. Notably, the text-augmented visual-text matching score leverages a $P$-score and an $N$-score in addition to the typical visual-text matching score. The $P$-score is utilized to close the visual-text domain gap through a surrogate captioning model, where the score is computed between the surrogate model-generated texts and the referring expression. The $N$-score considers the fine-grained alignment of region-text pairs via negative phrase mining, encouraging the masked image to be repelled from the mined distracting phrases. Extensive experiments are conducted on various datasets, including RefCOCO, RefCOCO+, and RefCOCOg. The proposed method clearly outperforms state-of-the-art zero-shot referring image segmentation methods. | [
"Suo, Yucheng",
"Zhu, Linchao",
"Yang, Yi"
] | Text Augmented Spatial Aware Zero-shot Referring Image Segmentation | findings-emnlp.73 | 2310.18049 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.74.bib | https://aclanthology.org/2023.findings-emnlp.74/ | @inproceedings{yosef-etal-2023-irfl,
title = "{IRFL}: Image Recognition of Figurative Language",
author = "Yosef, Ron and
Bitton, Yonatan and
Shahaf, Dafna",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.74",
doi = "10.18653/v1/2023.findings-emnlp.74",
pages = "1044--1058",
abstract = "Figures of speech such as metaphors, similes, and idioms are integral parts of human communication. They are ubiquitous in many forms of discourse, allowing people to convey complex, abstract ideas and evoke emotion. As figurative forms are often conveyed through multiple modalities (e.g., both text and images), understanding multimodal figurative language is an important AI challenge, weaving together profound vision, language, commonsense and cultural knowledge. In this work, we develop the Image Recognition of Figurative Language (IRFL) dataset. We leverage human annotation and an automatic pipeline we created to generate a multimodal dataset, and introduce two novel tasks as a benchmark for multimodal figurative language understanding. We experimented with state-of-the-art vision and language models and found that the best (22{\%}) performed substantially worse than humans (97{\%}). We release our dataset, benchmark, and code in hopes of driving the development of models that can better understand figurative language.",
}
| Figures of speech such as metaphors, similes, and idioms are integral parts of human communication. They are ubiquitous in many forms of discourse, allowing people to convey complex, abstract ideas and evoke emotion. As figurative forms are often conveyed through multiple modalities (e.g., both text and images), understanding multimodal figurative language is an important AI challenge, weaving together profound vision, language, commonsense and cultural knowledge. In this work, we develop the Image Recognition of Figurative Language (IRFL) dataset. We leverage human annotation and an automatic pipeline we created to generate a multimodal dataset, and introduce two novel tasks as a benchmark for multimodal figurative language understanding. We experimented with state-of-the-art vision and language models and found that the best (22{\%}) performed substantially worse than humans (97{\%}). We release our dataset, benchmark, and code in hopes of driving the development of models that can better understand figurative language. | [
"Yosef, Ron",
"Bitton, Yonatan",
"Shahaf, Dafna"
] | IRFL: Image Recognition of Figurative Language | findings-emnlp.74 | 2303.15445 | [
"https://github.com/irfl-dataset/irfl"
] | https://huggingface.co/papers/2303.15445 | 0 | 0 | 0 | 3 | [] | [
"lampent/IRFL",
"ColumbiaNLP/V-FLUTE"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.75.bib | https://aclanthology.org/2023.findings-emnlp.75/ | @inproceedings{pan-etal-2023-self,
title = "Self-supervised Meta-Prompt Learning with Meta-Gradient Regularization for Few-shot Generalization",
author = "Pan, Kaihang and
Li, Juncheng and
Song, Hongye and
Lin, Jun and
Liu, Xiaozhong and
Tang, Siliang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.75",
doi = "10.18653/v1/2023.findings-emnlp.75",
pages = "1059--1077",
abstract = "Prompt tuning is a parameter-efficient method, which learns soft prompts and conditions frozen language models to perform specific downstream tasks. Though effective, prompt tuning under few-shot settings on the one hand heavily relies on a good initialization of soft prompts. On the other hand, it can easily overfit to few-shot training samples, thereby undermining generalizability. Existing works leverage pre-training or supervised meta-learning to initialize soft prompts but they fail to data-efficiently generalize to unseen downstream tasks. To address the above problems, this paper proposes a novel Self-sUpervised meta-Prompt learning framework with MEta-gradient Regularization for few-shot generalization (SUPMER). SUPMER leverages self-supervised meta-learning with a diverse set of well-designed meta-tasks to learn a universal prompt initialization for efficient adaptation using only unlabeled data. Additionally, it jointly meta-learns a gradient regularization function to transform raw gradients into a domain-generalizable direction, thus alleviating the problem of overfitting. Extensive experiments show that SUPMER achieves better performance for different few-shot downstream tasks, and also exhibits a stronger domain generalization ability. The code for SUPMER will be available at https://github.com/beepkh/SUPMER.",
}
| Prompt tuning is a parameter-efficient method, which learns soft prompts and conditions frozen language models to perform specific downstream tasks. Though effective, prompt tuning under few-shot settings on the one hand heavily relies on a good initialization of soft prompts. On the other hand, it can easily overfit to few-shot training samples, thereby undermining generalizability. Existing works leverage pre-training or supervised meta-learning to initialize soft prompts but they fail to data-efficiently generalize to unseen downstream tasks. To address the above problems, this paper proposes a novel Self-sUpervised meta-Prompt learning framework with MEta-gradient Regularization for few-shot generalization (SUPMER). SUPMER leverages self-supervised meta-learning with a diverse set of well-designed meta-tasks to learn a universal prompt initialization for efficient adaptation using only unlabeled data. Additionally, it jointly meta-learns a gradient regularization function to transform raw gradients into a domain-generalizable direction, thus alleviating the problem of overfitting. Extensive experiments show that SUPMER achieves better performance for different few-shot downstream tasks, and also exhibits a stronger domain generalization ability. The code for SUPMER will be available at https://github.com/beepkh/SUPMER. | [
"Pan, Kaihang",
"Li, Juncheng",
"Song, Hongye",
"Lin, Jun",
"Liu, Xiaozhong",
"Tang, Siliang"
] | Self-supervised Meta-Prompt Learning with Meta-Gradient Regularization for Few-shot Generalization | findings-emnlp.75 | 2303.12314 | [
"https://github.com/beepkh/supmer"
] | https://huggingface.co/papers/2303.12314 | 0 | 1 | 0 | 6 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.76.bib | https://aclanthology.org/2023.findings-emnlp.76/ | @inproceedings{gao-etal-2023-adaptive,
title = "An Adaptive Prompt Generation Framework for Task-oriented Dialogue System",
author = "Gao, Jun and
Xiang, Liuyu and
Wu, Huijia and
Zhao, Han and
Tong, Yiqi and
He, Zhaofeng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.76",
doi = "10.18653/v1/2023.findings-emnlp.76",
pages = "1078--1089",
abstract = "The de facto way of utilizing black-box large language models (LLMs) to perform various downstream tasks is prompting. However, obtaining suitable prompts for specific tasks is still a challenging problem. While existing LLM-based methods demonstrate promising performance in task-oriented dialogue (TOD) task, they often require manual adjustment in prompt selection, or focus solely on dialogue understanding or generation. To address these issues, we propose an adaptive prompt generation framework to fully unleash the potential of LLMs for the comprehensive TOD system. Firstly, we design a trainable slot generator (TSG) that can generate domain and slot information in the belief state, which serves as prior knowledge for subsequent prompt generation. Next, we propose an adaptive prompt generator (APG) that utilizes the prior knowledge to generate prompts for the LLM, deriving the belief state and system response of the dialogue for evaluation. Finally, we evaluate our framework on the MultiWOZ 2.0 dataset. Extensive experiments demonstrate that our method outperforms existing methods. Our code and data will be released.",
}
| The de facto way of utilizing black-box large language models (LLMs) to perform various downstream tasks is prompting. However, obtaining suitable prompts for specific tasks is still a challenging problem. While existing LLM-based methods demonstrate promising performance in task-oriented dialogue (TOD) task, they often require manual adjustment in prompt selection, or focus solely on dialogue understanding or generation. To address these issues, we propose an adaptive prompt generation framework to fully unleash the potential of LLMs for the comprehensive TOD system. Firstly, we design a trainable slot generator (TSG) that can generate domain and slot information in the belief state, which serves as prior knowledge for subsequent prompt generation. Next, we propose an adaptive prompt generator (APG) that utilizes the prior knowledge to generate prompts for the LLM, deriving the belief state and system response of the dialogue for evaluation. Finally, we evaluate our framework on the MultiWOZ 2.0 dataset. Extensive experiments demonstrate that our method outperforms existing methods. Our code and data will be released. | [
"Gao, Jun",
"Xiang, Liuyu",
"Wu, Huijia",
"Zhao, Han",
"Tong, Yiqi",
"He, Zhaofeng"
] | An Adaptive Prompt Generation Framework for Task-oriented Dialogue System | findings-emnlp.76 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.77.bib | https://aclanthology.org/2023.findings-emnlp.77/ | @inproceedings{hou-etal-2023-temporal,
title = "Temporal Knowledge Graph Reasoning Based on N-tuple Modeling",
author = "Hou, Zhongni and
Jin, Xiaolong and
Li, Zixuan and
Bai, Long and
Guan, Saiping and
Zeng, Yutao and
Guo, Jiafeng and
Cheng, Xueqi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.77",
doi = "10.18653/v1/2023.findings-emnlp.77",
pages = "1090--1100",
abstract = "Reasoning over Temporal Knowledge Graphs (TKGs) that predicts temporal facts (e.g., events) in the future is crucial for many applications. The temporal facts in existing TKGs only contain their core entities (i.e., the entities playing core roles therein) and formulate them as quadruples, i.e., (subject entity, predicate, object entity, timestamp). This formulation oversimplifies temporal facts and inevitably causes information loss. Therefore, we propose to describe a temporal fact more accurately as an n-tuple, containing not only its predicate and core entities, but also its auxiliary entities, as well as the roles of all entities. By so doing, TKGs are augmented to N-tuple Temporal Knowledge Graphs (N-TKGs). To conduct reasoning over N-TKGs, we further propose N-tuple Evolutional Network (NE-Net). It recurrently learns the evolutional representations of entities and predicates in temporal facts at different timestamps in the history via modeling the relations among those entities and predicates. Based on the learned representations, reasoning tasks at future timestamps can be realized via task-specific decoders. Experiment results on two newly built datasets demonstrate the superiority of N-TKG and the effectiveness of NE-Net.",
}
| Reasoning over Temporal Knowledge Graphs (TKGs) that predicts temporal facts (e.g., events) in the future is crucial for many applications. The temporal facts in existing TKGs only contain their core entities (i.e., the entities playing core roles therein) and formulate them as quadruples, i.e., (subject entity, predicate, object entity, timestamp). This formulation oversimplifies temporal facts and inevitably causes information loss. Therefore, we propose to describe a temporal fact more accurately as an n-tuple, containing not only its predicate and core entities, but also its auxiliary entities, as well as the roles of all entities. By so doing, TKGs are augmented to N-tuple Temporal Knowledge Graphs (N-TKGs). To conduct reasoning over N-TKGs, we further propose N-tuple Evolutional Network (NE-Net). It recurrently learns the evolutional representations of entities and predicates in temporal facts at different timestamps in the history via modeling the relations among those entities and predicates. Based on the learned representations, reasoning tasks at future timestamps can be realized via task-specific decoders. Experiment results on two newly built datasets demonstrate the superiority of N-TKG and the effectiveness of NE-Net. | [
"Hou, Zhongni",
"Jin, Xiaolong",
"Li, Zixuan",
"Bai, Long",
"Guan, Saiping",
"Zeng, Yutao",
"Guo, Jiafeng",
"Cheng, Xueqi"
] | Temporal Knowledge Graph Reasoning Based on N-tuple Modeling | findings-emnlp.77 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.78.bib | https://aclanthology.org/2023.findings-emnlp.78/ | @inproceedings{du-etal-2023-make,
title = "Make Your Decision Convincing! A Unified Two-Stage Framework: Self-Attribution and Decision-Making",
author = "Du, Yanrui and
Zhao, Sendong and
Wang, Haochun and
Chen, Yuhan and
Bai, Rui and
Qiang, Zewen and
Cai, Muzhen and
Qin, Bing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.78",
doi = "10.18653/v1/2023.findings-emnlp.78",
pages = "1101--1112",
abstract = "Explaining black-box model behavior with natural language has achieved impressive results in various NLP tasks. Recent research has explored the utilization of subsequences from the input text as a rationale, providing users with evidence to support the model decision. Although existing frameworks excel in generating high-quality rationales while achieving high task performance, they neglect to account for the unreliable link between the generated rationale and model decision. In simpler terms, a model may make correct decisions while attributing wrong rationales, or make poor decisions while attributing correct rationales. To mitigate this issue, we propose a unified two-stage framework known as Self-Attribution and Decision-Making (SADM). Through extensive experiments on five reasoning datasets from the ERASER benchmark, we demonstrate that our framework not only establishes a more reliable link between the generated rationale and model decision but also achieves competitive results in task performance and the quality of rationale. Furthermore, we explore the potential of our framework in semi-supervised scenarios.",
}
| Explaining black-box model behavior with natural language has achieved impressive results in various NLP tasks. Recent research has explored the utilization of subsequences from the input text as a rationale, providing users with evidence to support the model decision. Although existing frameworks excel in generating high-quality rationales while achieving high task performance, they neglect to account for the unreliable link between the generated rationale and model decision. In simpler terms, a model may make correct decisions while attributing wrong rationales, or make poor decisions while attributing correct rationales. To mitigate this issue, we propose a unified two-stage framework known as Self-Attribution and Decision-Making (SADM). Through extensive experiments on five reasoning datasets from the ERASER benchmark, we demonstrate that our framework not only establishes a more reliable link between the generated rationale and model decision but also achieves competitive results in task performance and the quality of rationale. Furthermore, we explore the potential of our framework in semi-supervised scenarios. | [
"Du, Yanrui",
"Zhao, Sendong",
"Wang, Haochun",
"Chen, Yuhan",
"Bai, Rui",
"Qiang, Zewen",
"Cai, Muzhen",
"Qin, Bing"
] | Make Your Decision Convincing! A Unified Two-Stage Framework: Self-Attribution and Decision-Making | findings-emnlp.78 | 2310.13610 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.79.bib | https://aclanthology.org/2023.findings-emnlp.79/ | @inproceedings{niu-etal-2023-adaptive,
title = "Adaptive Structure Induction for Aspect-based Sentiment Analysis with Spectral Perspective",
author = "Niu, Hao and
Xiong, Yun and
Wang, Xiaosu and
Yu, Wenjing and
Zhang, Yao and
Guo, Zhonglei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.79",
doi = "10.18653/v1/2023.findings-emnlp.79",
pages = "1113--1126",
abstract = "Recently, incorporating structure information (e.g. dependency syntactic tree) can enhance the performance of aspect-based sentiment analysis (ABSA). However, this structure information is obtained from off-the-shelf parsers, which is often sub-optimal and cumbersome. Thus, automatically learning adaptive structures is conducive to solving this problem. In this work, we concentrate on structure induction from pre-trained language models (PLMs) and throw the structure induction into a spectrum perspective to explore the impact of scale information in language representation on structure induction ability. Concretely, the main architecture of our model is composed of commonly used PLMs (e.g. RoBERTa, etc), and a simple yet effective graph structure learning (GSL) module (graph learner + GNNs). Subsequently, we plug in spectral filters with different bands respectively after the PLMs to produce filtered language representations and feed them into the GSL module to induce latent structures. We conduct extensive experiments on three public benchmarks for ABSA. The results and further analyses demonstrate that introducing this spectral approach can shorten Aspects-sentiment Distance (AsD) and be beneficial to structure induction. Even based on such a simple framework, the effects on three datasets can reach SOTA (state of the art) or near SOTA performance. Additionally, our exploration also has the potential to be generalized to other tasks or to bring inspiration to other similar domains.",
}
| Recently, incorporating structure information (e.g. dependency syntactic tree) can enhance the performance of aspect-based sentiment analysis (ABSA). However, this structure information is obtained from off-the-shelf parsers, which is often sub-optimal and cumbersome. Thus, automatically learning adaptive structures is conducive to solving this problem. In this work, we concentrate on structure induction from pre-trained language models (PLMs) and throw the structure induction into a spectrum perspective to explore the impact of scale information in language representation on structure induction ability. Concretely, the main architecture of our model is composed of commonly used PLMs (e.g. RoBERTa, etc), and a simple yet effective graph structure learning (GSL) module (graph learner + GNNs). Subsequently, we plug in spectral filters with different bands respectively after the PLMs to produce filtered language representations and feed them into the GSL module to induce latent structures. We conduct extensive experiments on three public benchmarks for ABSA. The results and further analyses demonstrate that introducing this spectral approach can shorten Aspects-sentiment Distance (AsD) and be beneficial to structure induction. Even based on such a simple framework, the effects on three datasets can reach SOTA (state of the art) or near SOTA performance. Additionally, our exploration also has the potential to be generalized to other tasks or to bring inspiration to other similar domains. | [
"Niu, Hao",
"Xiong, Yun",
"Wang, Xiaosu",
"Yu, Wenjing",
"Zhang, Yao",
"Guo, Zhonglei"
] | Adaptive Structure Induction for Aspect-based Sentiment Analysis with Spectral Perspective | findings-emnlp.79 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.80.bib | https://aclanthology.org/2023.findings-emnlp.80/ | @inproceedings{west-etal-2023-novacomet,
title = "{N}ova{COMET}: Open Commonsense Foundation Models with Symbolic Knowledge Distillation",
author = "West, Peter and
Bras, Ronan and
Sorensen, Taylor and
Lin, Bill and
Jiang, Liwei and
Lu, Ximing and
Chandu, Khyathi and
Hessel, Jack and
Baheti, Ashutosh and
Bhagavatula, Chandra and
Choi, Yejin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.80",
doi = "10.18653/v1/2023.findings-emnlp.80",
pages = "1127--1149",
abstract = "We present NovaCOMET, an open commonsense knowledge model, that combines the best aspects of knowledge and general task models. Compared to previous knowledge models, NovaCOMET allows open-format relations enabling direct application to reasoning tasks; compared to general task models like Flan-T5, it explicitly centers knowledge, enabling superior performance for commonsense reasoning. NovaCOMET leverages the knowledge of opaque proprietary models to create an open knowledge pipeline. First, knowledge is symbolically distilled into NovATOMIC, a publicly-releaseddiscrete knowledge graph which can be audited, critiqued, and filtered. Next, we train NovaCOMET on NovATOMIC by fine-tuning an open-source pretrained model. NovaCOMET uses an open-format training objective, replacing the fixed relation sets of past knowledge models, enabling arbitrary structures within the data to serve as inputs or outputs. The resulting generation model, optionally augmented with human annotation, matches or exceeds comparable open task models like Flan-T5 on a range of commonsense generation tasks. NovaCOMET serves as a counterexample to the contemporary focus on instruction tuning only, demonstrating a distinct advantage to explicitly modeling commonsense knowledge as well.",
}
| We present NovaCOMET, an open commonsense knowledge model, that combines the best aspects of knowledge and general task models. Compared to previous knowledge models, NovaCOMET allows open-format relations enabling direct application to reasoning tasks; compared to general task models like Flan-T5, it explicitly centers knowledge, enabling superior performance for commonsense reasoning. NovaCOMET leverages the knowledge of opaque proprietary models to create an open knowledge pipeline. First, knowledge is symbolically distilled into NovATOMIC, a publicly-releaseddiscrete knowledge graph which can be audited, critiqued, and filtered. Next, we train NovaCOMET on NovATOMIC by fine-tuning an open-source pretrained model. NovaCOMET uses an open-format training objective, replacing the fixed relation sets of past knowledge models, enabling arbitrary structures within the data to serve as inputs or outputs. The resulting generation model, optionally augmented with human annotation, matches or exceeds comparable open task models like Flan-T5 on a range of commonsense generation tasks. NovaCOMET serves as a counterexample to the contemporary focus on instruction tuning only, demonstrating a distinct advantage to explicitly modeling commonsense knowledge as well. | [
"West, Peter",
"Bras, Ronan",
"Sorensen, Taylor",
"Lin, Bill",
"Jiang, Liwei",
"Lu, Ximing",
"Ch",
"u, Khyathi",
"Hessel, Jack",
"Baheti, Ashutosh",
"Bhagavatula, Ch",
"ra",
"Choi, Yejin"
] | NovaCOMET: Open Commonsense Foundation Models with Symbolic Knowledge Distillation | findings-emnlp.80 | 2312.05979 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.81.bib | https://aclanthology.org/2023.findings-emnlp.81/ | @inproceedings{iter-etal-2023-context,
title = "In-Context Demonstration Selection with Cross Entropy Difference",
author = "Iter, Dan and
Pryzant, Reid and
Xu, Ruochen and
Wang, Shuohang and
Liu, Yang and
Xu, Yichong and
Zhu, Chenguang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.81",
doi = "10.18653/v1/2023.findings-emnlp.81",
pages = "1150--1162",
abstract = "Large language models (LLMs) can use in-context demonstrations to improve performance on zero-shot tasks. However, selecting the best in-context examples is challenging because model performance can vary widely depending on the selected examples. We present a cross-entropy difference (CED) method for selecting in-context demonstrations. Our method is based on the observation that the effectiveness of in-context demonstrations negatively correlates with the perplexity of the test example by a language model that was finetuned on that demonstration. We utilize parameter efficient finetuning to train small models on training data that are used for computing the cross-entropy difference between a test example and every candidate in-context demonstration. This metric is used to rank and select in-context demonstrations independently for each test input. We evaluate our method on a mix-domain dataset that combines 8 benchmarks, representing 4 text generation tasks, showing that CED for in-context demonstration selection can improve performance for a variety of LLMs over baseline selection methods.",
}
| Large language models (LLMs) can use in-context demonstrations to improve performance on zero-shot tasks. However, selecting the best in-context examples is challenging because model performance can vary widely depending on the selected examples. We present a cross-entropy difference (CED) method for selecting in-context demonstrations. Our method is based on the observation that the effectiveness of in-context demonstrations negatively correlates with the perplexity of the test example by a language model that was finetuned on that demonstration. We utilize parameter efficient finetuning to train small models on training data that are used for computing the cross-entropy difference between a test example and every candidate in-context demonstration. This metric is used to rank and select in-context demonstrations independently for each test input. We evaluate our method on a mix-domain dataset that combines 8 benchmarks, representing 4 text generation tasks, showing that CED for in-context demonstration selection can improve performance for a variety of LLMs over baseline selection methods. | [
"Iter, Dan",
"Pryzant, Reid",
"Xu, Ruochen",
"Wang, Shuohang",
"Liu, Yang",
"Xu, Yichong",
"Zhu, Chenguang"
] | In-Context Demonstration Selection with Cross Entropy Difference | findings-emnlp.81 | 2305.14726 | [
"https://github.com/microsoft/lmops"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.82.bib | https://aclanthology.org/2023.findings-emnlp.82/ | @inproceedings{baylor-etal-2023-past,
title = "The Past, Present, and Future of Typological Databases in {NLP}",
author = "Baylor, Emi and
Ploeger, Esther and
Bjerva, Johannes",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.82",
doi = "10.18653/v1/2023.findings-emnlp.82",
pages = "1163--1169",
abstract = "Typological information has the potential to be beneficial in the development of NLP models, particularly for low-resource languages. Unfortunately, current large-scale typological databases, notably WALS and Grambank, are inconsistent both with each other and with other sources of typological information, such as linguistic grammars. Some of these inconsistencies stem from coding errors or linguistic variation, but many of the disagreements are due to the discrete categorical nature of these databases. We shed light on this issue by systematically exploring disagreements across typological databases and resources, and their uses in NLP, covering the past and present. We next investigate the future of such work, offering an argument that a continuous view of typological features is clearly beneficial, echoing recommendations from linguistics. We propose that such a view of typology has significant potential in the future, including in language modeling in low-resource scenarios.",
}
| Typological information has the potential to be beneficial in the development of NLP models, particularly for low-resource languages. Unfortunately, current large-scale typological databases, notably WALS and Grambank, are inconsistent both with each other and with other sources of typological information, such as linguistic grammars. Some of these inconsistencies stem from coding errors or linguistic variation, but many of the disagreements are due to the discrete categorical nature of these databases. We shed light on this issue by systematically exploring disagreements across typological databases and resources, and their uses in NLP, covering the past and present. We next investigate the future of such work, offering an argument that a continuous view of typological features is clearly beneficial, echoing recommendations from linguistics. We propose that such a view of typology has significant potential in the future, including in language modeling in low-resource scenarios. | [
"Baylor, Emi",
"Ploeger, Esther",
"Bjerva, Johannes"
] | The Past, Present, and Future of Typological Databases in NLP | findings-emnlp.82 | 2310.13440 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.83.bib | https://aclanthology.org/2023.findings-emnlp.83/ | @inproceedings{chen-etal-2023-soulchat,
title = "{S}oul{C}hat: Improving {LLM}s{'} Empathy, Listening, and Comfort Abilities through Fine-tuning with Multi-turn Empathy Conversations",
author = "Chen, Yirong and
Xing, Xiaofen and
Lin, Jingkai and
Zheng, Huimin and
Wang, Zhenyu and
Liu, Qi and
Xu, Xiangmin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.83",
doi = "10.18653/v1/2023.findings-emnlp.83",
pages = "1170--1183",
abstract = "Large language models (LLMs) have been widely applied in various fields due to their excellent capability for memorizing knowledge and chain of thought (CoT). When these language models are applied in the field of psychological counseling, they often rush to provide universal advice. However, when users seek psychological support, they need to gain empathy, trust, understanding and comfort, rather than just reasonable advice. To this end, we constructed a multi-turn empathetic conversation dataset of more than 2 million samples, in which the input is the multi-turn conversation context, and the target is empathetic responses that cover expressions such as questioning, comfort, recognition, listening, trust, emotional support, etc. Experiments have shown that the empathy ability of LLMs can be significantly enhanced when finetuning by using multi-turn dialogue history and responses that are closer to the expression of a psychological consultant.",
}
| Large language models (LLMs) have been widely applied in various fields due to their excellent capability for memorizing knowledge and chain of thought (CoT). When these language models are applied in the field of psychological counseling, they often rush to provide universal advice. However, when users seek psychological support, they need to gain empathy, trust, understanding and comfort, rather than just reasonable advice. To this end, we constructed a multi-turn empathetic conversation dataset of more than 2 million samples, in which the input is the multi-turn conversation context, and the target is empathetic responses that cover expressions such as questioning, comfort, recognition, listening, trust, emotional support, etc. Experiments have shown that the empathy ability of LLMs can be significantly enhanced when finetuning by using multi-turn dialogue history and responses that are closer to the expression of a psychological consultant. | [
"Chen, Yirong",
"Xing, Xiaofen",
"Lin, Jingkai",
"Zheng, Huimin",
"Wang, Zhenyu",
"Liu, Qi",
"Xu, Xiangmin"
] | SoulChat: Improving LLMs' Empathy, Listening, and Comfort Abilities through Fine-tuning with Multi-turn Empathy Conversations | findings-emnlp.83 | 2311.00273 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.84.bib | https://aclanthology.org/2023.findings-emnlp.84/ | @inproceedings{rao-etal-2023-chatgpt,
title = "Can {C}hat{GPT} Assess Human Personalities? A General Evaluation Framework",
author = "Rao, Haocong and
Leung, Cyril and
Miao, Chunyan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.84",
doi = "10.18653/v1/2023.findings-emnlp.84",
pages = "1184--1194",
abstract = "Large Language Models (LLMs) especially ChatGPT have produced impressive results in various areas, but their potential human-like psychology is still largely unexplored. Existing works study the virtual personalities of LLMs but rarely explore the possibility of analyzing human personalities via LLMs. This paper presents a generic evaluation framework for LLMs to assess human personalities based on Myers{--}Briggs Type Indicator (MBTI) tests. Specifically, we first devise unbiased prompts by randomly permuting options in MBTI questions and adopt the average testing result to encourage more impartial answer generation. Then, we propose to replace the subject in question statements to enable flexible queries and assessments on different subjects from LLMs. Finally, we re-formulate the question instructions in a manner of correctness evaluation to facilitate LLMs to generate clearer responses. The proposed framework enables LLMs to flexibly assess personalities of different groups of people. We further propose three evaluation metrics to measure the consistency, robustness, and fairness of assessment results from state-of-the-art LLMs including ChatGPT and GPT-4. Our experiments reveal ChatGPT{'}s ability to assess human personalities, and the average results demonstrate that it can achieve more consistent and fairer assessments in spite of lower robustness against prompt biases compared with InstructGPT.",
}
| Large Language Models (LLMs) especially ChatGPT have produced impressive results in various areas, but their potential human-like psychology is still largely unexplored. Existing works study the virtual personalities of LLMs but rarely explore the possibility of analyzing human personalities via LLMs. This paper presents a generic evaluation framework for LLMs to assess human personalities based on Myers{--}Briggs Type Indicator (MBTI) tests. Specifically, we first devise unbiased prompts by randomly permuting options in MBTI questions and adopt the average testing result to encourage more impartial answer generation. Then, we propose to replace the subject in question statements to enable flexible queries and assessments on different subjects from LLMs. Finally, we re-formulate the question instructions in a manner of correctness evaluation to facilitate LLMs to generate clearer responses. The proposed framework enables LLMs to flexibly assess personalities of different groups of people. We further propose three evaluation metrics to measure the consistency, robustness, and fairness of assessment results from state-of-the-art LLMs including ChatGPT and GPT-4. Our experiments reveal ChatGPT{'}s ability to assess human personalities, and the average results demonstrate that it can achieve more consistent and fairer assessments in spite of lower robustness against prompt biases compared with InstructGPT. | [
"Rao, Haocong",
"Leung, Cyril",
"Miao, Chunyan"
] | Can ChatGPT Assess Human Personalities? A General Evaluation Framework | findings-emnlp.84 | 2303.01248 | [
"https://github.com/kali-hac/chatgpt-mbti"
] | https://huggingface.co/papers/2303.01248 | 0 | 1 | 0 | 3 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.85.bib | https://aclanthology.org/2023.findings-emnlp.85/ | @inproceedings{zhang-etal-2023-moqagpt,
title = "{M}oqa{GPT} : Zero-Shot Multi-modal Open-domain Question Answering with Large Language Model",
author = "Zhang, Le and
Wu, Yihong and
Mo, Fengran and
Nie, Jian-Yun and
Agrawal, Aishwarya",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.85",
doi = "10.18653/v1/2023.findings-emnlp.85",
pages = "1195--1210",
abstract = "Multi-modal open-domain question answering typically requires evidence retrieval from databases across diverse modalities, such as images, tables, passages, etc. Even Large Language Models (LLMs) like GPT-4 fall short in this task. To enable LLMs to tackle the task in a zero-shot manner, we introduce MoqaGPT, a straightforward and flexible framework. Using a divide-and-conquer strategy that bypasses intricate multi-modality ranking, our framework can accommodate new modalities and seamlessly transition to new models for the task. Built upon LLMs, MoqaGPT retrieves and extracts answers from each modality separately, then fuses this multi-modal information using LLMs to produce a final answer. Our methodology boosts performance on the MMCoQA dataset, improving F1 by +37.91 points and EM by +34.07 points over the supervised baseline. On the MultiModalQA dataset, MoqaGPT surpasses the zero-shot baseline, improving F1 by 9.5 points and EM by 10.1 points, and significantly closes the gap with supervised methods. Our codebase is available at https://github.com/lezhang7/MOQAGPT.",
}
| Multi-modal open-domain question answering typically requires evidence retrieval from databases across diverse modalities, such as images, tables, passages, etc. Even Large Language Models (LLMs) like GPT-4 fall short in this task. To enable LLMs to tackle the task in a zero-shot manner, we introduce MoqaGPT, a straightforward and flexible framework. Using a divide-and-conquer strategy that bypasses intricate multi-modality ranking, our framework can accommodate new modalities and seamlessly transition to new models for the task. Built upon LLMs, MoqaGPT retrieves and extracts answers from each modality separately, then fuses this multi-modal information using LLMs to produce a final answer. Our methodology boosts performance on the MMCoQA dataset, improving F1 by +37.91 points and EM by +34.07 points over the supervised baseline. On the MultiModalQA dataset, MoqaGPT surpasses the zero-shot baseline, improving F1 by 9.5 points and EM by 10.1 points, and significantly closes the gap with supervised methods. Our codebase is available at https://github.com/lezhang7/MOQAGPT. | [
"Zhang, Le",
"Wu, Yihong",
"Mo, Fengran",
"Nie, Jian-Yun",
"Agrawal, Aishwarya"
] | MoqaGPT : Zero-Shot Multi-modal Open-domain Question Answering with Large Language Model | findings-emnlp.85 | 2310.13265 | [
"https://github.com/lezhang7/moqagpt"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.86.bib | https://aclanthology.org/2023.findings-emnlp.86/ | @inproceedings{mao-etal-2023-large,
title = "Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search",
author = "Mao, Kelong and
Dou, Zhicheng and
Mo, Fengran and
Hou, Jiewen and
Chen, Haonan and
Qian, Hongjin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.86",
doi = "10.18653/v1/2023.findings-emnlp.86",
pages = "1211--1225",
abstract = "Precisely understanding users{'} contextual search intent has been an important challenge for conversational search. As conversational search sessions are much more diverse and long-tailed, existing methods trained on limited data still show unsatisfactory effectiveness and robustness to handle real conversational search scenarios. Recently, large language models (LLMs) have demonstrated amazing capabilities for text generation and conversation understanding. In this work, we present a simple yet effective prompting framework, called LLM4CS, to leverage LLMs as a text-based search intent interpreter to help conversational search. Under this framework, we explore three prompting methods to generate multiple query rewrites and hypothetical responses, and propose to aggregate them into an integrated representation that can robustly represent the user{'}s real contextual search intent. Extensive automatic evaluations and human evaluations on three widely used conversational search benchmarks, including CAsT-19, CAsT-20, and CAsT-21, demonstrate the remarkable performance of our simple LLM4CS framework compared with existing methods and even using human rewrites. Our findings provide important evidence to better understand and leverage LLMs for conversational search.",
}
| Precisely understanding users{'} contextual search intent has been an important challenge for conversational search. As conversational search sessions are much more diverse and long-tailed, existing methods trained on limited data still show unsatisfactory effectiveness and robustness to handle real conversational search scenarios. Recently, large language models (LLMs) have demonstrated amazing capabilities for text generation and conversation understanding. In this work, we present a simple yet effective prompting framework, called LLM4CS, to leverage LLMs as a text-based search intent interpreter to help conversational search. Under this framework, we explore three prompting methods to generate multiple query rewrites and hypothetical responses, and propose to aggregate them into an integrated representation that can robustly represent the user{'}s real contextual search intent. Extensive automatic evaluations and human evaluations on three widely used conversational search benchmarks, including CAsT-19, CAsT-20, and CAsT-21, demonstrate the remarkable performance of our simple LLM4CS framework compared with existing methods and even using human rewrites. Our findings provide important evidence to better understand and leverage LLMs for conversational search. | [
"Mao, Kelong",
"Dou, Zhicheng",
"Mo, Fengran",
"Hou, Jiewen",
"Chen, Haonan",
"Qian, Hongjin"
] | Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search | findings-emnlp.86 | 2303.06573 | [
"https://github.com/kyriemao/llmcs"
] | https://huggingface.co/papers/2303.06573 | 0 | 0 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.87.bib | https://aclanthology.org/2023.findings-emnlp.87/ | @inproceedings{bao-etal-2023-docasref,
title = "{D}oc{A}s{R}ef: An Empirical Study on Repurposing Reference-based Summary Quality Metrics as Reference-free Metrics",
author = "Bao, Forrest and
Tu, Ruixuan and
Luo, Ge and
Yang, Yinfei and
Li, Hebi and
Qiu, Minghui and
He, Youbiao and
Chen, Cen",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.87",
doi = "10.18653/v1/2023.findings-emnlp.87",
pages = "1226--1235",
abstract = "Automated summary quality assessment falls into two categories: reference-based and reference-free. Reference-based metrics, historically deemed more accurate due to the additional information provided by human-written references, are limited by their reliance on human input. In this paper, we hypothesize that the comparison methodologies used by some reference-based metrics to evaluate a system summary against its corresponding reference can be effectively adapted to assess it against its source document, thereby transforming these metrics into reference-free ones. Experimental results support this hypothesis. After being repurposed reference-freely, the zero-shot BERTScore using the pretrained DeBERTa-large-MNLI model of $<$0.5B parameters consistently outperforms its original reference-based version across various aspects on the SummEval and Newsroom datasets. It also excels in comparison to most existing reference-free metrics and closely competes with zero-shot summary evaluators based on GPT-3.5.",
}
| Automated summary quality assessment falls into two categories: reference-based and reference-free. Reference-based metrics, historically deemed more accurate due to the additional information provided by human-written references, are limited by their reliance on human input. In this paper, we hypothesize that the comparison methodologies used by some reference-based metrics to evaluate a system summary against its corresponding reference can be effectively adapted to assess it against its source document, thereby transforming these metrics into reference-free ones. Experimental results support this hypothesis. After being repurposed reference-freely, the zero-shot BERTScore using the pretrained DeBERTa-large-MNLI model of $<$0.5B parameters consistently outperforms its original reference-based version across various aspects on the SummEval and Newsroom datasets. It also excels in comparison to most existing reference-free metrics and closely competes with zero-shot summary evaluators based on GPT-3.5. | [
"Bao, Forrest",
"Tu, Ruixuan",
"Luo, Ge",
"Yang, Yinfei",
"Li, Hebi",
"Qiu, Minghui",
"He, Youbiao",
"Chen, Cen"
] | DocAsRef: An Empirical Study on Repurposing Reference-based Summary Quality Metrics as Reference-free Metrics | findings-emnlp.87 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.88.bib | https://aclanthology.org/2023.findings-emnlp.88/ | @inproceedings{deshpande-etal-2023-toxicity,
title = "Toxicity in chatgpt: Analyzing persona-assigned language models",
author = "Deshpande, Ameet and
Murahari, Vishvak and
Rajpurohit, Tanmay and
Kalyan, Ashwin and
Narasimhan, Karthik",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.88",
doi = "10.18653/v1/2023.findings-emnlp.88",
pages = "1236--1270",
abstract = "Large language models (LLMs) have shown incredible capabilities and transcended the natural language processing (NLP) community, with adoption throughout many services like healthcare, therapy, education, and customer service. Since users include people with critical information needs like students or patients engaging with chatbots, the safety of these systems is of prime importance. Legislation has recognized its significance and recently drafted a {``}Blueprint For An AI Bill Of Rights{''} which calls for domain experts to identify risks and potential impact of AI systems. To this end, we systematically evaluate toxicity in over half a million generations of ChatGPT, a popular dialogue-based LLM. We find that setting the system parameter of ChatGPT by assigning it a persona, say that of the boxer Muhammad Ali, significantly increases the toxicity of generations. Depending on the persona assigned to ChatGPT, its toxicity can increase up to $6\times$, with outputs engaging in incorrect stereotypes, harmful dialogue, and hurtful opinions. Furthermore, we find concerning patterns where specific entities (e.g., certain races) are targeted more than others ($3\times$ more) irrespective of the assigned persona, reflecting discriminatory biases in the model. Our findings show that multiple provisions in the legislative blueprint are being violated, and we hope that the broader AI community rethinks the efficacy of current safety guardrails and develops better techniques that lead to robust, safe, and trustworthy AI.",
}
| Large language models (LLMs) have shown incredible capabilities and transcended the natural language processing (NLP) community, with adoption throughout many services like healthcare, therapy, education, and customer service. Since users include people with critical information needs like students or patients engaging with chatbots, the safety of these systems is of prime importance. Legislation has recognized its significance and recently drafted a {``}Blueprint For An AI Bill Of Rights{''} which calls for domain experts to identify risks and potential impact of AI systems. To this end, we systematically evaluate toxicity in over half a million generations of ChatGPT, a popular dialogue-based LLM. We find that setting the system parameter of ChatGPT by assigning it a persona, say that of the boxer Muhammad Ali, significantly increases the toxicity of generations. Depending on the persona assigned to ChatGPT, its toxicity can increase up to $6\times$, with outputs engaging in incorrect stereotypes, harmful dialogue, and hurtful opinions. Furthermore, we find concerning patterns where specific entities (e.g., certain races) are targeted more than others ($3\times$ more) irrespective of the assigned persona, reflecting discriminatory biases in the model. Our findings show that multiple provisions in the legislative blueprint are being violated, and we hope that the broader AI community rethinks the efficacy of current safety guardrails and develops better techniques that lead to robust, safe, and trustworthy AI. | [
"Deshp",
"e, Ameet",
"Murahari, Vishvak",
"Rajpurohit, Tanmay",
"Kalyan, Ashwin",
"Narasimhan, Karthik"
] | Toxicity in chatgpt: Analyzing persona-assigned language models | findings-emnlp.88 | 2304.05335 | [
""
] | https://huggingface.co/papers/2304.05335 | 0 | 0 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.89.bib | https://aclanthology.org/2023.findings-emnlp.89/ | @inproceedings{wang-etal-2023-execution,
title = "Execution-Based Evaluation for Open-Domain Code Generation",
author = "Wang, Zhiruo and
Zhou, Shuyan and
Fried, Daniel and
Neubig, Graham",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.89",
doi = "10.18653/v1/2023.findings-emnlp.89",
pages = "1271--1290",
abstract = "To extend the scope of coding queries to more realistic settings, we propose ODEX, the first Open-Domain EXecution-based natural language (NL) to Python code generation dataset. ODEX has 945 NL-Code pairs spanning 79 diverse libraries, along with 1,707 human-written test cases for execution. Our NL-Code pairs are harvested from StackOverflow forums to encourage natural and practical coding queries. Moreover, ODEX supports four natural languages as intents, in English, Spanish, Japanese, and Russian. ODEX unveils intriguing behavioral differences among top-performing code language models (LM). While CODEX achieves better overall results, CODEGEN improves effectively via scaling {--} CODEGEN 6.1B performs comparably with CODEX 12B. Both models show substantial gaps between open and closed domains, but CODEGEN gaps tend to decrease with model size while CODEX gaps increase. We release ODEX to facilitate research into open-domain problems for the code generation community.",
}
| To extend the scope of coding queries to more realistic settings, we propose ODEX, the first Open-Domain EXecution-based natural language (NL) to Python code generation dataset. ODEX has 945 NL-Code pairs spanning 79 diverse libraries, along with 1,707 human-written test cases for execution. Our NL-Code pairs are harvested from StackOverflow forums to encourage natural and practical coding queries. Moreover, ODEX supports four natural languages as intents, in English, Spanish, Japanese, and Russian. ODEX unveils intriguing behavioral differences among top-performing code language models (LM). While CODEX achieves better overall results, CODEGEN improves effectively via scaling {--} CODEGEN 6.1B performs comparably with CODEX 12B. Both models show substantial gaps between open and closed domains, but CODEGEN gaps tend to decrease with model size while CODEX gaps increase. We release ODEX to facilitate research into open-domain problems for the code generation community. | [
"Wang, Zhiruo",
"Zhou, Shuyan",
"Fried, Daniel",
"Neubig, Graham"
] | Execution-Based Evaluation for Open-Domain Code Generation | findings-emnlp.89 | 2212.10481 | [
"https://github.com/zorazrw/odex"
] | https://huggingface.co/papers/2212.10481 | 0 | 1 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.90.bib | https://aclanthology.org/2023.findings-emnlp.90/ | @inproceedings{zhang-etal-2023-syntax,
title = "Syntax-Aware Retrieval Augmented Code Generation",
author = "Zhang, Xiangyu and
Zhou, Yu and
Yang, Guang and
Chen, Taolue",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.90",
doi = "10.18653/v1/2023.findings-emnlp.90",
pages = "1291--1302",
abstract = "Neural code generation models are nowadays widely adopted to generate code from natural language descriptions automatically. Recently, pre-trained neural models equipped with token-level retrieval capabilities have exhibited great potentials in neural machine translation. However, applying them directly to code generation experience challenges: the use of the retrieval-based mechanism inevitably introduces extraneous noise to the generation process, resulting in even syntactically incorrect code. Computationally, such models necessitate frequent searches of the cached datastore, which turns out to be time-consuming. To address these issues, we propose $k$NN-TRANX, a token-level retrieval augmented code generation method. $k$NN-TRANX allows for searches in smaller datastores tailored for the code generation task. It leverages syntax constraints for the retrieval of datastores, which reduces the impact of retrieve noise. We evaluate $k$NN-TRANX on two public datasets and the experimental results confirm the effectiveness of our approach.",
}
| Neural code generation models are nowadays widely adopted to generate code from natural language descriptions automatically. Recently, pre-trained neural models equipped with token-level retrieval capabilities have exhibited great potentials in neural machine translation. However, applying them directly to code generation experience challenges: the use of the retrieval-based mechanism inevitably introduces extraneous noise to the generation process, resulting in even syntactically incorrect code. Computationally, such models necessitate frequent searches of the cached datastore, which turns out to be time-consuming. To address these issues, we propose $k$NN-TRANX, a token-level retrieval augmented code generation method. $k$NN-TRANX allows for searches in smaller datastores tailored for the code generation task. It leverages syntax constraints for the retrieval of datastores, which reduces the impact of retrieve noise. We evaluate $k$NN-TRANX on two public datasets and the experimental results confirm the effectiveness of our approach. | [
"Zhang, Xiangyu",
"Zhou, Yu",
"Yang, Guang",
"Chen, Taolue"
] | Syntax-Aware Retrieval Augmented Code Generation | findings-emnlp.90 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.91.bib | https://aclanthology.org/2023.findings-emnlp.91/ | @inproceedings{sui-etal-2023-selecting,
title = "Selecting Key Views for Zero-Shot Entity Linking",
author = "Sui, Xuhui and
Zhang, Ying and
Song, Kehui and
Zhou, Baohang and
Yuan, Xiaojie and
Zhang, Wensheng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.91",
doi = "10.18653/v1/2023.findings-emnlp.91",
pages = "1303--1312",
abstract = "Entity linking, which aligns mentions in the text to entities in knowledge bases, is essential for many natural language processing tasks. Considering the real-world scenarios, recent research hotspot of entity linking has focused on the zero-shot setting, where mentions need to link to unseen entities and only the description of each entity is provided. This task challenges the language understanding ability of models to capture the coherence evidence between the mention context and entity description. However, entity descriptions often contain rich information from multiple views, and a mention with context only relates to a small part of the information. Other irrelevant information will introduce noise, which interferes with models to make the right judgments. Furthermore, the existence of these information also makes it difficult to synthesize key information. To solve these problems, we select key views from descriptions and propose a KVZEL framework for zero-shot entity linking. Specifically, our KVZEL first adopts unsupervised clustering to form sub views. Then, it employs a mention-aware key views selection module to iteratively accumulate mention-focused views. This puts emphasis on capturing mention-related information and allows long-range key information integration. Finally, we aggregate key views to make the final decision. Experimental results show the effectiveness of our KVZEL and it achieves the new state-of-the-art on the zero-shot entity linking dataset.",
}
| Entity linking, which aligns mentions in the text to entities in knowledge bases, is essential for many natural language processing tasks. Considering the real-world scenarios, recent research hotspot of entity linking has focused on the zero-shot setting, where mentions need to link to unseen entities and only the description of each entity is provided. This task challenges the language understanding ability of models to capture the coherence evidence between the mention context and entity description. However, entity descriptions often contain rich information from multiple views, and a mention with context only relates to a small part of the information. Other irrelevant information will introduce noise, which interferes with models to make the right judgments. Furthermore, the existence of these information also makes it difficult to synthesize key information. To solve these problems, we select key views from descriptions and propose a KVZEL framework for zero-shot entity linking. Specifically, our KVZEL first adopts unsupervised clustering to form sub views. Then, it employs a mention-aware key views selection module to iteratively accumulate mention-focused views. This puts emphasis on capturing mention-related information and allows long-range key information integration. Finally, we aggregate key views to make the final decision. Experimental results show the effectiveness of our KVZEL and it achieves the new state-of-the-art on the zero-shot entity linking dataset. | [
"Sui, Xuhui",
"Zhang, Ying",
"Song, Kehui",
"Zhou, Baohang",
"Yuan, Xiaojie",
"Zhang, Wensheng"
] | Selecting Key Views for Zero-Shot Entity Linking | findings-emnlp.91 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.92.bib | https://aclanthology.org/2023.findings-emnlp.92/ | @inproceedings{hsu-etal-2023-explanation,
title = "Is Explanation the Cure? Misinformation Mitigation in the Short Term and Long Term",
author = "Hsu, Yi-Li and
Dai, Shih-Chieh and
Xiong, Aiping and
Ku, Lun-Wei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.92",
doi = "10.18653/v1/2023.findings-emnlp.92",
pages = "1313--1323",
abstract = "With advancements in natural language processing (NLP) models, automatic explanation generation has been proposed to mitigate misinformation on social media platforms in addition to adding warning labels to identified fake news. While many researchers have focused on generating good explanations, how these explanations can really help humans combat fake news is under-explored. In this study, we compare the effectiveness of a warning label and the state-of- the-art counterfactual explanations generated by GPT-4 in debunking misinformation. In a two-wave, online human-subject study, participants (N = 215) were randomly assigned to a control group in which false contents are shown without any intervention, a warning tag group in which the false claims were labeled, or an explanation group in which the false contents were accompanied by GPT-4 generated explanations. Our results show that both interventions significantly decrease participants{'} self-reported belief in fake claims in an equivalent manner for the short-term and long-term. We discuss the implications of our findings and directions for future NLP-based misinformation debunking strategies.",
}
| With advancements in natural language processing (NLP) models, automatic explanation generation has been proposed to mitigate misinformation on social media platforms in addition to adding warning labels to identified fake news. While many researchers have focused on generating good explanations, how these explanations can really help humans combat fake news is under-explored. In this study, we compare the effectiveness of a warning label and the state-of- the-art counterfactual explanations generated by GPT-4 in debunking misinformation. In a two-wave, online human-subject study, participants (N = 215) were randomly assigned to a control group in which false contents are shown without any intervention, a warning tag group in which the false claims were labeled, or an explanation group in which the false contents were accompanied by GPT-4 generated explanations. Our results show that both interventions significantly decrease participants{'} self-reported belief in fake claims in an equivalent manner for the short-term and long-term. We discuss the implications of our findings and directions for future NLP-based misinformation debunking strategies. | [
"Hsu, Yi-Li",
"Dai, Shih-Chieh",
"Xiong, Aiping",
"Ku, Lun-Wei"
] | Is Explanation the Cure? Misinformation Mitigation in the Short Term and Long Term | findings-emnlp.92 | 2310.17711 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.93.bib | https://aclanthology.org/2023.findings-emnlp.93/ | @inproceedings{krishna-etal-2023-improving,
title = "Improving the Robustness of Summarization Models by Detecting and Removing Input Noise",
author = "Krishna, Kundan and
Zhao, Yao and
Ren, Jie and
Lakshminarayanan, Balaji and
Luo, Jiaming and
Saleh, Mohammad and
Liu, Peter",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.93",
doi = "10.18653/v1/2023.findings-emnlp.93",
pages = "1324--1336",
abstract = "The evaluation of abstractive summarization models typically uses test data that is identically distributed as training data. In real-world practice, documents to be summarized may contain input noise caused by text extraction artifacts or data pipeline bugs. The robustness of model performance under distribution shift caused by such noise is relatively under studied. We present a large empirical study quantifying the sometimes severe loss in performance {--} up to 12 ROUGE-1 points {--} from different types of input noise for a range of datasets and model sizes. We then propose a light-weight method for detecting and removing such noise in the input during model inference without requiring any extra training, auxiliary models, or even prior knowledge of the type of noise. Our proposed approach effectively mitigates the loss in performance, recovering a large fraction of the performance drop, sometimes as large as 11 ROUGE-1 points.",
}
| The evaluation of abstractive summarization models typically uses test data that is identically distributed as training data. In real-world practice, documents to be summarized may contain input noise caused by text extraction artifacts or data pipeline bugs. The robustness of model performance under distribution shift caused by such noise is relatively under studied. We present a large empirical study quantifying the sometimes severe loss in performance {--} up to 12 ROUGE-1 points {--} from different types of input noise for a range of datasets and model sizes. We then propose a light-weight method for detecting and removing such noise in the input during model inference without requiring any extra training, auxiliary models, or even prior knowledge of the type of noise. Our proposed approach effectively mitigates the loss in performance, recovering a large fraction of the performance drop, sometimes as large as 11 ROUGE-1 points. | [
"Krishna, Kundan",
"Zhao, Yao",
"Ren, Jie",
"Lakshminarayanan, Balaji",
"Luo, Jiaming",
"Saleh, Mohammad",
"Liu, Peter"
] | Improving the Robustness of Summarization Models by Detecting and Removing Input Noise | findings-emnlp.93 | 2212.09928 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.94.bib | https://aclanthology.org/2023.findings-emnlp.94/ | @inproceedings{kumarage-etal-2023-reliable,
title = "How Reliable Are {AI}-Generated-Text Detectors? An Assessment Framework Using Evasive Soft Prompts",
author = "Kumarage, Tharindu and
Sheth, Paras and
Moraffah, Raha and
Garland, Joshua and
Liu, Huan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.94",
doi = "10.18653/v1/2023.findings-emnlp.94",
pages = "1337--1349",
abstract = "In recent years, there has been a rapid proliferation of AI-generated text, primarily driven by the release of powerful pre-trained language models (PLMs). To address the issue of misuse associated with AI-generated text, various high-performing detectors have been developed, including the OpenAI detector and the Stanford DetectGPT. In our study, we ask how reliable these detectors are. We answer the question by designing a novel approach that can prompt any PLM to generate text that evades these high-performing detectors. The proposed approach suggests a universal evasive prompt, a novel type of soft prompt, which guides PLMs in producing {``}human-like{''} text that can mislead the detectors. The novel universal evasive prompt is achieved in two steps: First, we create an evasive soft prompt tailored to a specific PLM through prompt tuning; and then, we leverage the transferability of soft prompts to transfer the learned evasive soft prompt from one PLM to another. Employing multiple PLMs in various writing tasks, we conduct extensive experiments to evaluate the efficacy of the evasive soft prompts in their evasion of state-of-the-art detectors.",
}
| In recent years, there has been a rapid proliferation of AI-generated text, primarily driven by the release of powerful pre-trained language models (PLMs). To address the issue of misuse associated with AI-generated text, various high-performing detectors have been developed, including the OpenAI detector and the Stanford DetectGPT. In our study, we ask how reliable these detectors are. We answer the question by designing a novel approach that can prompt any PLM to generate text that evades these high-performing detectors. The proposed approach suggests a universal evasive prompt, a novel type of soft prompt, which guides PLMs in producing {``}human-like{''} text that can mislead the detectors. The novel universal evasive prompt is achieved in two steps: First, we create an evasive soft prompt tailored to a specific PLM through prompt tuning; and then, we leverage the transferability of soft prompts to transfer the learned evasive soft prompt from one PLM to another. Employing multiple PLMs in various writing tasks, we conduct extensive experiments to evaluate the efficacy of the evasive soft prompts in their evasion of state-of-the-art detectors. | [
"Kumarage, Tharindu",
"Sheth, Paras",
"Moraffah, Raha",
"Garl",
", Joshua",
"Liu, Huan"
] | How Reliable Are AI-Generated-Text Detectors? An Assessment Framework Using Evasive Soft Prompts | findings-emnlp.94 | 2310.05095 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.95.bib | https://aclanthology.org/2023.findings-emnlp.95/ | @inproceedings{gueta-etal-2023-knowledge,
title = "Knowledge is a Region in Weight Space for Fine-tuned Language Models",
author = "Gueta, Almog and
Venezian, Elad and
Raffel, Colin and
Slonim, Noam and
Katz, Yoav and
Choshen, Leshem",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.95",
doi = "10.18653/v1/2023.findings-emnlp.95",
pages = "1350--1370",
abstract = "Research on neural networks has focused on understanding a single model trained on a single dataset. However, relatively little is known about the relationships between different models, particularly those trained or tested on different datasets. We address this by studying how the weight space and the underlying loss landscape of different models are interconnected. Specifically, we demonstrate that finetuned models that were optimized for high performance, reside in well-defined regions in weight space, and vice versa {--} that any model that resides anywhere in those regions also exhibits high performance. Notably, we show that language models that have been finetuned on the same dataset form a tight cluster in the weight space, while models finetuned on different datasets from the same underlying task form a looser cluster. Moreover, traversing around the region between the models leads to new models that perform comparably or even better than models obtained via finetuning, even on tasks that the original models were not finetuned on. Our findings provide insight into the relationships between models, demonstrating that a model positioned between two similar models can acquire the knowledge of both. We leverage this and design a method for selecting a better model for efficient finetuning. Specifically, we show that starting from the center of the region is as effective, if not more, than using the pretrained model in 11 out of 12 datasets, resulting in an average accuracy improvement of 3.06.",
}
| Research on neural networks has focused on understanding a single model trained on a single dataset. However, relatively little is known about the relationships between different models, particularly those trained or tested on different datasets. We address this by studying how the weight space and the underlying loss landscape of different models are interconnected. Specifically, we demonstrate that finetuned models that were optimized for high performance, reside in well-defined regions in weight space, and vice versa {--} that any model that resides anywhere in those regions also exhibits high performance. Notably, we show that language models that have been finetuned on the same dataset form a tight cluster in the weight space, while models finetuned on different datasets from the same underlying task form a looser cluster. Moreover, traversing around the region between the models leads to new models that perform comparably or even better than models obtained via finetuning, even on tasks that the original models were not finetuned on. Our findings provide insight into the relationships between models, demonstrating that a model positioned between two similar models can acquire the knowledge of both. We leverage this and design a method for selecting a better model for efficient finetuning. Specifically, we show that starting from the center of the region is as effective, if not more, than using the pretrained model in 11 out of 12 datasets, resulting in an average accuracy improvement of 3.06. | [
"Gueta, Almog",
"Venezian, Elad",
"Raffel, Colin",
"Slonim, Noam",
"Katz, Yoav",
"Choshen, Leshem"
] | Knowledge is a Region in Weight Space for Fine-tuned Language Models | findings-emnlp.95 | 2302.04863 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.96.bib | https://aclanthology.org/2023.findings-emnlp.96/ | @inproceedings{kadasi-singh-2023-unveiling,
title = "Unveiling the Multi-Annotation Process: Examining the Influence of Annotation Quantity and Instance Difficulty on Model Performance",
author = "Kadasi, Pritam and
Singh, Mayank",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.96",
doi = "10.18653/v1/2023.findings-emnlp.96",
pages = "1371--1388",
abstract = "The NLP community has long advocated for the construction of multi-annotator datasets to better capture the nuances of language interpretation, subjectivity, and ambiguity. This paper conducts a retrospective study to show how performance scores can vary when a dataset expands from a single annotation per instance to multiple annotations. We propose a novel multi-annotator simulation process to generate datasets with varying annotation budgets. We show that similar datasets with the same annotation budget can lead to varying performance gains. Our findings challenge the popular belief that models trained on multi-annotation examples always lead to better performance than models trained on single or few-annotation examples.",
}
| The NLP community has long advocated for the construction of multi-annotator datasets to better capture the nuances of language interpretation, subjectivity, and ambiguity. This paper conducts a retrospective study to show how performance scores can vary when a dataset expands from a single annotation per instance to multiple annotations. We propose a novel multi-annotator simulation process to generate datasets with varying annotation budgets. We show that similar datasets with the same annotation budget can lead to varying performance gains. Our findings challenge the popular belief that models trained on multi-annotation examples always lead to better performance than models trained on single or few-annotation examples. | [
"Kadasi, Pritam",
"Singh, Mayank"
] | Unveiling the Multi-Annotation Process: Examining the Influence of Annotation Quantity and Instance Difficulty on Model Performance | findings-emnlp.96 | 2310.14572 | [
""
] | https://huggingface.co/papers/2310.14572 | 0 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.97.bib | https://aclanthology.org/2023.findings-emnlp.97/ | @inproceedings{pan-etal-2023-risk,
title = "On the Risk of Misinformation Pollution with Large Language Models",
author = "Pan, Yikang and
Pan, Liangming and
Chen, Wenhu and
Nakov, Preslav and
Kan, Min-Yen and
Wang, William",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.97",
doi = "10.18653/v1/2023.findings-emnlp.97",
pages = "1389--1403",
abstract = "We investigate the potential misuse of modern Large Language Models (LLMs) for generating credible-sounding misinformation and its subsequent impact on information-intensive applications, particularly Open-Domain Question Answering (ODQA) systems. We establish a threat model and simulate potential misuse scenarios, both unintentional and intentional, to assess the extent to which LLMs can be utilized to produce misinformation. Our study reveals that LLMs can act as effective misinformation generators, leading to a significant degradation (up to 87{\%}) in the performance of ODQA systems. Moreover, we uncover disparities in the attributes associated with persuading humans and machines, presenting an obstacle to current human-centric approaches to combat misinformation. To mitigate the harm caused by LLM-generated misinformation, we propose three defense strategies: misinformation detection, vigilant prompting, and reader ensemble. These approaches have demonstrated promising results, albeit with certain associated costs. Lastly, we discuss the practicality of utilizing LLMs as automatic misinformation generators and provide relevant resources and code to facilitate future research in this area.",
}
| We investigate the potential misuse of modern Large Language Models (LLMs) for generating credible-sounding misinformation and its subsequent impact on information-intensive applications, particularly Open-Domain Question Answering (ODQA) systems. We establish a threat model and simulate potential misuse scenarios, both unintentional and intentional, to assess the extent to which LLMs can be utilized to produce misinformation. Our study reveals that LLMs can act as effective misinformation generators, leading to a significant degradation (up to 87{\%}) in the performance of ODQA systems. Moreover, we uncover disparities in the attributes associated with persuading humans and machines, presenting an obstacle to current human-centric approaches to combat misinformation. To mitigate the harm caused by LLM-generated misinformation, we propose three defense strategies: misinformation detection, vigilant prompting, and reader ensemble. These approaches have demonstrated promising results, albeit with certain associated costs. Lastly, we discuss the practicality of utilizing LLMs as automatic misinformation generators and provide relevant resources and code to facilitate future research in this area. | [
"Pan, Yikang",
"Pan, Liangming",
"Chen, Wenhu",
"Nakov, Preslav",
"Kan, Min-Yen",
"Wang, William"
] | On the Risk of Misinformation Pollution with Large Language Models | findings-emnlp.97 | 2305.13661 | [
"https://github.com/MexicanLemonade/LLM-Misinfo-QA"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.98.bib | https://aclanthology.org/2023.findings-emnlp.98/ | @inproceedings{nagoudi-etal-2023-dolphin,
title = "Dolphin: A Challenging and Diverse Benchmark for {A}rabic {NLG}",
author = "Nagoudi, El Moatez Billah and
Elmadany, AbdelRahim and
El-Shangiti, Ahmed and
Abdul-Mageed, Muhammad",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.98",
doi = "10.18653/v1/2023.findings-emnlp.98",
pages = "1404--1422",
abstract = "We present Dolphin, a novel benchmark that addresses the need for a natural language generation (NLG) evaluation framework dedicated to the wide collection of Arabic languages and varieties. The proposed benchmark encompasses a broad range of 13 different NLG tasks, including dialogue generation, question answering, machine translation, summarization, among others. Dolphin comprises a substantial corpus of 40 diverse and representative public datasets across 50 test splits, carefully curated to reflect real-world scenarios and the linguistic richness of Arabic. It sets a new standard for evaluating the performance and generalization capabilities of Arabic and multilingual models, promising to enable researchers to push the boundaries of current methodologies. We provide an extensive analysis of Dolphin, highlighting its diversity and identifying gaps in current Arabic NLG research. We also offer a public leaderboard that is both interactive and modular and evaluate several Arabic and multilingual models on our benchmark, allowing us to set strong baselines against which researchers can compare.",
}
| We present Dolphin, a novel benchmark that addresses the need for a natural language generation (NLG) evaluation framework dedicated to the wide collection of Arabic languages and varieties. The proposed benchmark encompasses a broad range of 13 different NLG tasks, including dialogue generation, question answering, machine translation, summarization, among others. Dolphin comprises a substantial corpus of 40 diverse and representative public datasets across 50 test splits, carefully curated to reflect real-world scenarios and the linguistic richness of Arabic. It sets a new standard for evaluating the performance and generalization capabilities of Arabic and multilingual models, promising to enable researchers to push the boundaries of current methodologies. We provide an extensive analysis of Dolphin, highlighting its diversity and identifying gaps in current Arabic NLG research. We also offer a public leaderboard that is both interactive and modular and evaluate several Arabic and multilingual models on our benchmark, allowing us to set strong baselines against which researchers can compare. | [
"Nagoudi, El Moatez Billah",
"Elmadany, AbdelRahim",
"El-Shangiti, Ahmed",
"Abdul-Mageed, Muhammad"
] | Dolphin: A Challenging and Diverse Benchmark for Arabic NLG | findings-emnlp.98 | 2305.14989 | [
""
] | https://huggingface.co/papers/2305.14989 | 2 | 0 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.99.bib | https://aclanthology.org/2023.findings-emnlp.99/ | @inproceedings{fu-etal-2023-hierarchical,
title = "Hierarchical Enhancement Framework for Aspect-based Argument Mining",
author = "Fu, Yujie and
Li, Yang and
Wang, Suge and
Li, Xiaoli and
Li, Deyu and
Liao, Jian and
Zheng, JianXing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.99",
doi = "10.18653/v1/2023.findings-emnlp.99",
pages = "1423--1433",
abstract = "Aspect-Based Argument Mining (ABAM) is a critical task in computational argumentation. Existing methods have primarily treated ABAM as a nested named entity recognition problem, overlooking the need for tailored strategies to effectively address the specific challenges of ABAM tasks. To this end, we propose a layer-based Hierarchical Enhancement Framework (HEF) for ABAM, and introduce three novel components: the Semantic and Syntactic Fusion (SSF) component, the Batch-level Heterogeneous Graph Attention Network (BHGAT) component, and the Span Mask Interactive Attention (SMIA) component. These components serve the purposes of optimizing underlying representations, detecting argument unit stances, and constraining aspect term recognition boundaries, respectively. By incorporating these components, our framework enables better handling of the challenges and improves the performance and accuracy in argument unit and aspect term recognition. Experiments on multiple datasets and various tasks verify the effectiveness of the proposed framework and components.",
}
| Aspect-Based Argument Mining (ABAM) is a critical task in computational argumentation. Existing methods have primarily treated ABAM as a nested named entity recognition problem, overlooking the need for tailored strategies to effectively address the specific challenges of ABAM tasks. To this end, we propose a layer-based Hierarchical Enhancement Framework (HEF) for ABAM, and introduce three novel components: the Semantic and Syntactic Fusion (SSF) component, the Batch-level Heterogeneous Graph Attention Network (BHGAT) component, and the Span Mask Interactive Attention (SMIA) component. These components serve the purposes of optimizing underlying representations, detecting argument unit stances, and constraining aspect term recognition boundaries, respectively. By incorporating these components, our framework enables better handling of the challenges and improves the performance and accuracy in argument unit and aspect term recognition. Experiments on multiple datasets and various tasks verify the effectiveness of the proposed framework and components. | [
"Fu, Yujie",
"Li, Yang",
"Wang, Suge",
"Li, Xiaoli",
"Li, Deyu",
"Liao, Jian",
"Zheng, JianXing"
] | Hierarchical Enhancement Framework for Aspect-based Argument Mining | findings-emnlp.99 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.100.bib | https://aclanthology.org/2023.findings-emnlp.100/ | @inproceedings{wei-etal-2023-menatqa,
title = "{M}enat{QA}: A New Dataset for Testing the Temporal Comprehension and Reasoning Abilities of Large Language Models",
author = "Wei, Yifan and
Su, Yisong and
Ma, Huanhuan and
Yu, Xiaoyan and
Lei, Fangyu and
Zhang, Yuanzhe and
Zhao, Jun and
Liu, Kang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.100",
doi = "10.18653/v1/2023.findings-emnlp.100",
pages = "1434--1447",
abstract = "Large language models (LLMs) have shown nearly saturated performance on many natural language processing (NLP) tasks. As a result, it is natural for people to believe that LLMs have also mastered abilities such as time understanding and reasoning. However, research on the temporal sensitivity of LLMs has been insufficiently emphasized. To fill this gap, this paper constructs Multiple Sensitive Factors Time QA (MenatQA), which encompasses three temporal factors (scope factor, order factor, counterfactual factor) with total 2,853 samples for evaluating the time comprehension and reasoning abilities of LLMs. This paper tests current mainstream LLMs with different parameter sizes, ranging from billions to hundreds of billions. The results show most LLMs fall behind smaller temporal reasoning models with different degree on these factors. In specific, LLMs show a significant vulnerability to temporal biases and depend heavily on the temporal information provided in questions. Furthermore, this paper undertakes a preliminary investigation into potential improvement strategies by devising specific prompts and leveraging external tools. These approaches serve as valuable baselines or references for future research endeavors.",
}
| Large language models (LLMs) have shown nearly saturated performance on many natural language processing (NLP) tasks. As a result, it is natural for people to believe that LLMs have also mastered abilities such as time understanding and reasoning. However, research on the temporal sensitivity of LLMs has been insufficiently emphasized. To fill this gap, this paper constructs Multiple Sensitive Factors Time QA (MenatQA), which encompasses three temporal factors (scope factor, order factor, counterfactual factor) with total 2,853 samples for evaluating the time comprehension and reasoning abilities of LLMs. This paper tests current mainstream LLMs with different parameter sizes, ranging from billions to hundreds of billions. The results show most LLMs fall behind smaller temporal reasoning models with different degree on these factors. In specific, LLMs show a significant vulnerability to temporal biases and depend heavily on the temporal information provided in questions. Furthermore, this paper undertakes a preliminary investigation into potential improvement strategies by devising specific prompts and leveraging external tools. These approaches serve as valuable baselines or references for future research endeavors. | [
"Wei, Yifan",
"Su, Yisong",
"Ma, Huanhuan",
"Yu, Xiaoyan",
"Lei, Fangyu",
"Zhang, Yuanzhe",
"Zhao, Jun",
"Liu, Kang"
] | MenatQA: A New Dataset for Testing the Temporal Comprehension and Reasoning Abilities of Large Language Models | findings-emnlp.100 | 2310.05157 | [
"https://github.com/weiyifan1023/MenatQA"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.101.bib | https://aclanthology.org/2023.findings-emnlp.101/ | @inproceedings{madaan-etal-2023-makes,
title = "What Makes Chain-of-Thought Prompting Effective? A Counterfactual Study",
author = "Madaan, Aman and
Hermann, Katherine and
Yazdanbakhsh, Amir",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.101",
doi = "10.18653/v1/2023.findings-emnlp.101",
pages = "1448--1535",
abstract = "The effectiveness of Chain-of-thought prompting (CoT) has been widely recognized, but the underlying mechanisms behind its success, the reason why it just works for a wide range of tasks, remains an open question. To investigate this, we employ a counterfactual prompting approach, systematically manipulating elements of examples used in a few-shot prompt, and testing the consequences on model behavior. This allows us to understand the relative contributions of prompt elements such as symbols (digits, entities) and patterns (equations, sentence structure) on in-context learning. Our experiments with three different large language models (LLMs) reveal several key findings. First, the specific symbols used in the prompt do not significantly impact the model{'}s performance. However, consistent patterns in examples and specifying text in style frequently found on the web are crucial. Second, our findings suggest that the necessity of accurate few-shot examples depends on their role in communicating task understanding. We identify tasks where inaccurate few-shot examples hurt and, surprisingly, tasks where they improve performance. Additionally, we find that the intermediate steps in CoT may not necessarily facilitate learning how to solve a task, but instead efficiently convey task understanding (what) to the model. Furthermore, CoT leverages LLMs to fill in missing commonsense information, particularly helping difficult reasoning problems and long-tail questions.",
}
| The effectiveness of Chain-of-thought prompting (CoT) has been widely recognized, but the underlying mechanisms behind its success, the reason why it just works for a wide range of tasks, remains an open question. To investigate this, we employ a counterfactual prompting approach, systematically manipulating elements of examples used in a few-shot prompt, and testing the consequences on model behavior. This allows us to understand the relative contributions of prompt elements such as symbols (digits, entities) and patterns (equations, sentence structure) on in-context learning. Our experiments with three different large language models (LLMs) reveal several key findings. First, the specific symbols used in the prompt do not significantly impact the model{'}s performance. However, consistent patterns in examples and specifying text in style frequently found on the web are crucial. Second, our findings suggest that the necessity of accurate few-shot examples depends on their role in communicating task understanding. We identify tasks where inaccurate few-shot examples hurt and, surprisingly, tasks where they improve performance. Additionally, we find that the intermediate steps in CoT may not necessarily facilitate learning how to solve a task, but instead efficiently convey task understanding (what) to the model. Furthermore, CoT leverages LLMs to fill in missing commonsense information, particularly helping difficult reasoning problems and long-tail questions. | [
"Madaan, Aman",
"Hermann, Katherine",
"Yazdanbakhsh, Amir"
] | What Makes Chain-of-Thought Prompting Effective? A Counterfactual Study | findings-emnlp.101 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.102.bib | https://aclanthology.org/2023.findings-emnlp.102/ | @inproceedings{loyola-etal-2023-perceptual,
title = "Perceptual Structure in the absence of grounding: the impact of abstractedness and subjectivity in color language for {LLM}s",
author = "Loyola, Pablo and
Marrese-Taylor, Edison and
Hoyos-Idrobo, Andres",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.102",
doi = "10.18653/v1/2023.findings-emnlp.102",
pages = "1536--1542",
abstract = "The need for grounding in language understanding is an active research topic. Previous work has suggested that color perception and color language appear as a suitable test bed to empirically study the problem, given its cognitive significance and showing that there is considerable alignment between a defined color space and the feature space defined by a language model. To further study this issue, we collect a large scale source of colors and their descriptions, containing almost a 1 million examples , and perform an empirical analysis to compare two kinds of alignments: (i) inter-space, by learning a mapping between embedding space and color space, and (ii) intra-space, by means of prompting comparatives between color descriptions. Our results show that while color space alignment holds for monolexemic, highly pragmatic color descriptions, this alignment drops considerably in the presence of examples that exhibit elements of real linguistic usage such as subjectivity and abstractedness, suggesting that grounding may be required in such cases.",
}
| The need for grounding in language understanding is an active research topic. Previous work has suggested that color perception and color language appear as a suitable test bed to empirically study the problem, given its cognitive significance and showing that there is considerable alignment between a defined color space and the feature space defined by a language model. To further study this issue, we collect a large scale source of colors and their descriptions, containing almost a 1 million examples , and perform an empirical analysis to compare two kinds of alignments: (i) inter-space, by learning a mapping between embedding space and color space, and (ii) intra-space, by means of prompting comparatives between color descriptions. Our results show that while color space alignment holds for monolexemic, highly pragmatic color descriptions, this alignment drops considerably in the presence of examples that exhibit elements of real linguistic usage such as subjectivity and abstractedness, suggesting that grounding may be required in such cases. | [
"Loyola, Pablo",
"Marrese-Taylor, Edison",
"Hoyos-Idrobo, Andres"
] | Perceptual Structure in the absence of grounding: the impact of abstractedness and subjectivity in color language for LLMs | findings-emnlp.102 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.103.bib | https://aclanthology.org/2023.findings-emnlp.103/ | @inproceedings{ihtiyar-etal-2023-dataset,
title = "A Dataset for Investigating the Impact of Context for Offensive Language Detection in Tweets",
author = {{\.I}htiyar, Musa and
{\"O}zdemir, {\"O}mer and
Ereng{\"u}l, Mustafa and
{\"O}zg{\"u}r, Arzucan},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.103",
doi = "10.18653/v1/2023.findings-emnlp.103",
pages = "1543--1549",
abstract = "Offensive language detection is crucial in natural language processing (NLP). We investigated the importance of context for detecting such language in reply tweets on Twitter, where the use of offensive language is widespread. We collected a Turkish tweet dataset where the target group was unvaccinated people during the Covid period. Tweets in the dataset were enriched with contextual information by adding the original tweet to which a particular tweet was posted as a reply. The dataset, which includes over 28,000 tweet-reply pairs, was manually labeled by human annotators and made publicly available. In addition, we compared the performance of different machine learning models with and without contextual information. Our results show that this type of contextual information was not very useful in improving the performance of the models in general, although it slightly increased the macro-averaged F1-score of certain models.",
}
| Offensive language detection is crucial in natural language processing (NLP). We investigated the importance of context for detecting such language in reply tweets on Twitter, where the use of offensive language is widespread. We collected a Turkish tweet dataset where the target group was unvaccinated people during the Covid period. Tweets in the dataset were enriched with contextual information by adding the original tweet to which a particular tweet was posted as a reply. The dataset, which includes over 28,000 tweet-reply pairs, was manually labeled by human annotators and made publicly available. In addition, we compared the performance of different machine learning models with and without contextual information. Our results show that this type of contextual information was not very useful in improving the performance of the models in general, although it slightly increased the macro-averaged F1-score of certain models. | [
"{\\.I}htiyar, Musa",
"{\\\"O}zdemir, {\\\"O}mer",
"Ereng{\\\"u}l, Mustafa",
"{\\\"O}zg{\\\"u}r, Arzucan"
] | A Dataset for Investigating the Impact of Context for Offensive Language Detection in Tweets | findings-emnlp.103 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.104.bib | https://aclanthology.org/2023.findings-emnlp.104/ | @inproceedings{ciosici-etal-2023-remember,
title = "Remember what you did so you know what to do next",
author = "Ciosici, Manuel and
Hedges, Alex and
Kankanampati, Yash and
Martin, Justin and
Freedman, Marjorie and
Weischedel, Ralph",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.104",
doi = "10.18653/v1/2023.findings-emnlp.104",
pages = "1550--1562",
abstract = "We explore using the 6B parameter GPT-J language model to create a plan for a simulated robot to achieve 30 classes of goals in ScienceWorld, a text game simulator for elementary science experiments and for which previously published empirical work has shown large language models (LLM)s to be a poor fit (Wang et al., 2022). Using the Markov assumption, the LLM outperforms the state-of-the-art based on reinforcement learning by a factor of 1.4. When we fill the LLM{'}s input buffer with as many prior steps as will fit, improvement rises to 3.3x. Even when training on only 6.5{\%} of the training data, we observe a 2.3x improvement over the state-of-the-art. Our experiments show that performance varies widely across the 30 classes of actions, indicating that averaging over tasks can hide significant performance issues.",
}
| We explore using the 6B parameter GPT-J language model to create a plan for a simulated robot to achieve 30 classes of goals in ScienceWorld, a text game simulator for elementary science experiments and for which previously published empirical work has shown large language models (LLM)s to be a poor fit (Wang et al., 2022). Using the Markov assumption, the LLM outperforms the state-of-the-art based on reinforcement learning by a factor of 1.4. When we fill the LLM{'}s input buffer with as many prior steps as will fit, improvement rises to 3.3x. Even when training on only 6.5{\%} of the training data, we observe a 2.3x improvement over the state-of-the-art. Our experiments show that performance varies widely across the 30 classes of actions, indicating that averaging over tasks can hide significant performance issues. | [
"Ciosici, Manuel",
"Hedges, Alex",
"Kankanampati, Yash",
"Martin, Justin",
"Freedman, Marjorie",
"Weischedel, Ralph"
] | Remember what you did so you know what to do next | findings-emnlp.104 | 2311.01468 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.105.bib | https://aclanthology.org/2023.findings-emnlp.105/ | @inproceedings{sung-etal-2023-empirical,
title = "An Empirical Study of Multimodal Model Merging",
author = "Sung, Yi-Lin and
Li, Linjie and
Lin, Kevin and
Gan, Zhe and
Bansal, Mohit and
Wang, Lijuan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.105",
doi = "10.18653/v1/2023.findings-emnlp.105",
pages = "1563--1575",
abstract = "Model merging (e.g., via interpolation or task arithmetic) fuses multiple models trained on different tasks to generate a multi-task solution. The technique has been proven successful in previous studies, where the models are trained on similar tasks and with the same initialization. In this paper, we expand on this concept to a multimodal setup by merging transformers trained on different modalities. Furthermore, we conduct our study for a novel goal where we can merge vision, language, and cross-modal transformers of a modality-specific architecture to create a parameter-efficient modality-agnostic architecture. Through comprehensive experiments, we systematically investigate the key factors impacting model performance after merging, including initialization, merging mechanisms, and model architectures. We also propose two metrics that assess the distance between weights to be merged and can serve as an indicator of the merging outcomes. Our analysis leads to an effective training recipe for matching the performance of the modality-agnostic baseline (i.e., pre-trained from scratch) via model merging. Our method also outperforms naive merging significantly on various tasks, with improvements of 3{\%} on VQA, 7{\%} on COCO retrieval, 25{\%} on NLVR2, 14{\%} on Flickr30k and 3{\%} on ADE20k.",
}
| Model merging (e.g., via interpolation or task arithmetic) fuses multiple models trained on different tasks to generate a multi-task solution. The technique has been proven successful in previous studies, where the models are trained on similar tasks and with the same initialization. In this paper, we expand on this concept to a multimodal setup by merging transformers trained on different modalities. Furthermore, we conduct our study for a novel goal where we can merge vision, language, and cross-modal transformers of a modality-specific architecture to create a parameter-efficient modality-agnostic architecture. Through comprehensive experiments, we systematically investigate the key factors impacting model performance after merging, including initialization, merging mechanisms, and model architectures. We also propose two metrics that assess the distance between weights to be merged and can serve as an indicator of the merging outcomes. Our analysis leads to an effective training recipe for matching the performance of the modality-agnostic baseline (i.e., pre-trained from scratch) via model merging. Our method also outperforms naive merging significantly on various tasks, with improvements of 3{\%} on VQA, 7{\%} on COCO retrieval, 25{\%} on NLVR2, 14{\%} on Flickr30k and 3{\%} on ADE20k. | [
"Sung, Yi-Lin",
"Li, Linjie",
"Lin, Kevin",
"Gan, Zhe",
"Bansal, Mohit",
"Wang, Lijuan"
] | An Empirical Study of Multimodal Model Merging | findings-emnlp.105 | 2304.14933 | [
"https://github.com/ylsung/vl-merging"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.106.bib | https://aclanthology.org/2023.findings-emnlp.106/ | @inproceedings{behjati-etal-2023-learning,
title = "Learning to Abstract with Nonparametric Variational Information Bottleneck",
author = "Behjati, Melika and
Fehr, Fabio and
Henderson, James",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.106",
doi = "10.18653/v1/2023.findings-emnlp.106",
pages = "1576--1586",
abstract = "Learned representations at the level of characters, sub-words, words, and sentences, have each contributed to advances in understanding different NLP tasks and linguistic phenomena. However, learning textual embeddings is costly as they are tokenization specific and require different models to be trained for each level of abstraction. We introduce a novel language representation model which can learn to compress to different levels of abstraction at different layers of the same model. We apply Nonparametric Variational Information Bottleneck (NVIB) to stacked Transformer self-attention layers in the encoder, which encourages an information-theoretic compression of the representations through the model. We find that the layers within the model correspond to increasing levels of abstraction and that their representations are more linguistically informed. Finally, we show that NVIB compression results in a model which is more robust to adversarial perturbations.",
}
| Learned representations at the level of characters, sub-words, words, and sentences, have each contributed to advances in understanding different NLP tasks and linguistic phenomena. However, learning textual embeddings is costly as they are tokenization specific and require different models to be trained for each level of abstraction. We introduce a novel language representation model which can learn to compress to different levels of abstraction at different layers of the same model. We apply Nonparametric Variational Information Bottleneck (NVIB) to stacked Transformer self-attention layers in the encoder, which encourages an information-theoretic compression of the representations through the model. We find that the layers within the model correspond to increasing levels of abstraction and that their representations are more linguistically informed. Finally, we show that NVIB compression results in a model which is more robust to adversarial perturbations. | [
"Behjati, Melika",
"Fehr, Fabio",
"Henderson, James"
] | Learning to Abstract with Nonparametric Variational Information Bottleneck | findings-emnlp.106 | 2310.17284 | [
"https://github.com/idiap/nvib"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.107.bib | https://aclanthology.org/2023.findings-emnlp.107/ | @inproceedings{chen-etal-2023-global,
title = "Global Structure Knowledge-Guided Relation Extraction Method for Visually-Rich Document",
author = "Chen, Xiangnan and
Xiao, Qian and
Li, Juncheng and
Dong, Duo and
Lin, Jun and
Liu, Xiaozhong and
Tang, Siliang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.107",
doi = "10.18653/v1/2023.findings-emnlp.107",
pages = "1587--1598",
abstract = "Visual Relation Extraction (VRE) is a powerful means of discovering relationships between entities within visually-rich documents. Existing methods often focus on manipulating entity features to find pairwise relations, yet neglect the more fundamental structural information that links disparate entity pairs together. The absence of global structure information may make the model struggle to learn long-range relations and easily predict conflicted results. To alleviate such limitations, we propose a GlObal Structure knowledge-guided relation Extraction (GOSE) framework. GOSE initiates by generating preliminary relation predictions on entity pairs extracted from a scanned image of the document. Subsequently, global structural knowledge is captured from the preceding iterative predictions, which are then incorporated into the representations of the entities. This {``}generate-capture-incorporate{''} cycle is repeated multiple times, allowing entity representations and global structure knowledge to be mutually reinforced. Extensive experiments validate that GOSE not only outperforms existing methods in the standard fine-tuning setting but also reveals superior cross-lingual learning capabilities; indeed, even yields stronger data-efficient performance in the low-resource setting.",
}
| Visual Relation Extraction (VRE) is a powerful means of discovering relationships between entities within visually-rich documents. Existing methods often focus on manipulating entity features to find pairwise relations, yet neglect the more fundamental structural information that links disparate entity pairs together. The absence of global structure information may make the model struggle to learn long-range relations and easily predict conflicted results. To alleviate such limitations, we propose a GlObal Structure knowledge-guided relation Extraction (GOSE) framework. GOSE initiates by generating preliminary relation predictions on entity pairs extracted from a scanned image of the document. Subsequently, global structural knowledge is captured from the preceding iterative predictions, which are then incorporated into the representations of the entities. This {``}generate-capture-incorporate{''} cycle is repeated multiple times, allowing entity representations and global structure knowledge to be mutually reinforced. Extensive experiments validate that GOSE not only outperforms existing methods in the standard fine-tuning setting but also reveals superior cross-lingual learning capabilities; indeed, even yields stronger data-efficient performance in the low-resource setting. | [
"Chen, Xiangnan",
"Xiao, Qian",
"Li, Juncheng",
"Dong, Duo",
"Lin, Jun",
"Liu, Xiaozhong",
"Tang, Siliang"
] | Global Structure Knowledge-Guided Relation Extraction Method for Visually-Rich Document | findings-emnlp.107 | 2305.13850 | [
"https://github.com/chenxn2020/gose"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.108.bib | https://aclanthology.org/2023.findings-emnlp.108/ | @inproceedings{lin-etal-2023-learning,
title = "Learning to Compose Representations of Different Encoder Layers towards Improving Compositional Generalization",
author = "Lin, Lei and
Li, Shuangtao and
Zheng, Yafang and
Fu, Biao and
Liu, Shan and
Chen, Yidong and
Shi, Xiaodong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.108",
doi = "10.18653/v1/2023.findings-emnlp.108",
pages = "1599--1614",
abstract = "Recent studies have shown that sequence-to-sequence (seq2seq) models struggle with compositional generalization (CG), i.e., the ability to systematically generalize to unseen compositions of seen components. There is mounting evidence that one of the reasons hindering CG is the representation of the encoder uppermost layer is entangled, i.e., the syntactic and semantic representations of sequences are entangled. However, we consider that the previously identified representation entanglement problem is not comprehensive enough. Additionally, we hypothesize that the source keys and values representations passing into different decoder layers are also entangled. Starting from this intuition, we propose CompoSition (\textbf{Compo}se \textbf{S}yntactic and Semant\textbf{i}c Representa\textbf{tion}s), an extension to seq2seq models which learns to compose representations of different encoder layers dynamically for different tasks, since recent studies reveal that the bottom layers of the Transformer encoder contain more syntactic information and the top ones contain more semantic information. Specifically, we introduce a \textit{composed layer} between the encoder and decoder to compose different encoder layers{'} representations to generate specific keys and values passing into different decoder layers. CompoSition achieves competitive results on two comprehensive and realistic benchmarks, which empirically demonstrates the effectiveness of our proposal. Codes are available at \url{https://github.com/thinkaboutzero/COMPOSITION}.",
}
| Recent studies have shown that sequence-to-sequence (seq2seq) models struggle with compositional generalization (CG), i.e., the ability to systematically generalize to unseen compositions of seen components. There is mounting evidence that one of the reasons hindering CG is the representation of the encoder uppermost layer is entangled, i.e., the syntactic and semantic representations of sequences are entangled. However, we consider that the previously identified representation entanglement problem is not comprehensive enough. Additionally, we hypothesize that the source keys and values representations passing into different decoder layers are also entangled. Starting from this intuition, we propose CompoSition (\textbf{Compo}se \textbf{S}yntactic and Semant\textbf{i}c Representa\textbf{tion}s), an extension to seq2seq models which learns to compose representations of different encoder layers dynamically for different tasks, since recent studies reveal that the bottom layers of the Transformer encoder contain more syntactic information and the top ones contain more semantic information. Specifically, we introduce a \textit{composed layer} between the encoder and decoder to compose different encoder layers{'} representations to generate specific keys and values passing into different decoder layers. CompoSition achieves competitive results on two comprehensive and realistic benchmarks, which empirically demonstrates the effectiveness of our proposal. Codes are available at \url{https://github.com/thinkaboutzero/COMPOSITION}. | [
"Lin, Lei",
"Li, Shuangtao",
"Zheng, Yafang",
"Fu, Biao",
"Liu, Shan",
"Chen, Yidong",
"Shi, Xiaodong"
] | Learning to Compose Representations of Different Encoder Layers towards Improving Compositional Generalization | findings-emnlp.108 | 2305.12169 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.109.bib | https://aclanthology.org/2023.findings-emnlp.109/ | @inproceedings{brahma-etal-2023-selectnoise,
title = "{S}elect{N}oise: Unsupervised Noise Injection to Enable Zero-Shot Machine Translation for Extremely Low-resource Languages",
author = "Brahma, Maharaj and
Maurya, Kaushal and
Desarkar, Maunendra",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.109",
doi = "10.18653/v1/2023.findings-emnlp.109",
pages = "1615--1629",
abstract = "In this work, we focus on the task of machine translation (MT) from extremely low-resource language (ELRLs) to English. The unavailability of parallel data, lack of representation from large multilingual pre-trained models, and limited monolingual data hinder the development of MT systems for ELRLs. However, many ELRLs often share lexical similarities with high-resource languages (HRLs) due to factors such as dialectical variations, geographical proximity, and language structure. We utilize this property to improve cross-lingual signals from closely related HRL to enable MT for ELRLs. Specifically, we propose a novel unsupervised approach, $\textit{SelectNoise}$, based on $\textit{selective candidate extraction}$ and $\textit{noise injection}$ to generate noisy HRLs training data. The noise injection acts as a regularizer, and the model trained with noisy data learns to handle lexical variations such as spelling, grammar, and vocabulary changes, leading to improved cross-lingual transfer to ELRLs. The selective candidates are extracted using BPE merge operations and edit operations, and noise injection is performed using greedy, top-p, and top-k sampling strategies. We evaluate the proposed model on 12 ELRLs from the FLORES-200 benchmark in a zero-shot setting across two language families. The proposed model outperformed all the strong baselines, demonstrating its efficacy. It has comparable performance with the supervised noise injection model. Our code and model are publicly available.",
}
| In this work, we focus on the task of machine translation (MT) from extremely low-resource language (ELRLs) to English. The unavailability of parallel data, lack of representation from large multilingual pre-trained models, and limited monolingual data hinder the development of MT systems for ELRLs. However, many ELRLs often share lexical similarities with high-resource languages (HRLs) due to factors such as dialectical variations, geographical proximity, and language structure. We utilize this property to improve cross-lingual signals from closely related HRL to enable MT for ELRLs. Specifically, we propose a novel unsupervised approach, $\textit{SelectNoise}$, based on $\textit{selective candidate extraction}$ and $\textit{noise injection}$ to generate noisy HRLs training data. The noise injection acts as a regularizer, and the model trained with noisy data learns to handle lexical variations such as spelling, grammar, and vocabulary changes, leading to improved cross-lingual transfer to ELRLs. The selective candidates are extracted using BPE merge operations and edit operations, and noise injection is performed using greedy, top-p, and top-k sampling strategies. We evaluate the proposed model on 12 ELRLs from the FLORES-200 benchmark in a zero-shot setting across two language families. The proposed model outperformed all the strong baselines, demonstrating its efficacy. It has comparable performance with the supervised noise injection model. Our code and model are publicly available. | [
"Brahma, Maharaj",
"Maurya, Kaushal",
"Desarkar, Maunendra"
] | SelectNoise: Unsupervised Noise Injection to Enable Zero-Shot Machine Translation for Extremely Low-resource Languages | findings-emnlp.109 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.110.bib | https://aclanthology.org/2023.findings-emnlp.110/ | @inproceedings{chen-etal-2023-breaking,
title = "Breaking Boundaries in Retrieval Systems: Unsupervised Domain Adaptation with Denoise-Finetuning",
author = "Chen, Che and
Yang, Ching and
Lin, Chun-Yi and
Kao, Hung-Yu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.110",
doi = "10.18653/v1/2023.findings-emnlp.110",
pages = "1630--1642",
abstract = "Dense retrieval models have exhibited remarkable effectiveness, but they rely on abundant labeled data and face challenges when applied to different domains. Previous domain adaptation methods have employed generative models to generate pseudo queries, creating pseudo datasets to enhance the performance of dense retrieval models. However, these approaches typically use unadapted rerank models, leading to potentially imprecise labels. In this paper, we demonstrate the significance of adapting the rerank model to the target domain prior to utilizing it for label generation. This adaptation process enables us to obtain more accurate labels, thereby improving the overall performance of the dense retrieval model. Additionally, by combining the adapted retrieval model with the adapted rerank model, we achieve significantly better domain adaptation results across three retrieval datasets. We release our code for future research.",
}
| Dense retrieval models have exhibited remarkable effectiveness, but they rely on abundant labeled data and face challenges when applied to different domains. Previous domain adaptation methods have employed generative models to generate pseudo queries, creating pseudo datasets to enhance the performance of dense retrieval models. However, these approaches typically use unadapted rerank models, leading to potentially imprecise labels. In this paper, we demonstrate the significance of adapting the rerank model to the target domain prior to utilizing it for label generation. This adaptation process enables us to obtain more accurate labels, thereby improving the overall performance of the dense retrieval model. Additionally, by combining the adapted retrieval model with the adapted rerank model, we achieve significantly better domain adaptation results across three retrieval datasets. We release our code for future research. | [
"Chen, Che",
"Yang, Ching",
"Lin, Chun-Yi",
"Kao, Hung-Yu"
] | Breaking Boundaries in Retrieval Systems: Unsupervised Domain Adaptation with Denoise-Finetuning | findings-emnlp.110 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.111.bib | https://aclanthology.org/2023.findings-emnlp.111/ | @inproceedings{zhang-etal-2023-exploring-cognitive,
title = "Exploring the Cognitive Knowledge Structure of Large Language Models: An Educational Diagnostic Assessment Approach",
author = "Zhang, Zheyuan and
Yu, Jifan and
Li, Juanzi and
Hou, Lei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.111",
doi = "10.18653/v1/2023.findings-emnlp.111",
pages = "1643--1650",
abstract = "Large Language Models (LLMs) have not only exhibited exceptional performance across various tasks, but also demonstrated sparks of intelligence. Recent studies have focused on assessing their capabilities on human exams and revealed their impressive competence in different domains. However, cognitive research on the overall knowledge structure of LLMs is still lacking. In this paper, based on educational diagnostic assessment method, we conduct an evaluation using MoocRadar, a meticulously annotated human test dataset based on Bloom Taxonomy. We aim to reveal the knowledge structures of LLMs and gain insights of their cognitive capabilities. This research emphasizes the significance of investigating LLMs{'} knowledge and understanding the disparate cognitive patterns of LLMs. By shedding light on models{'} knowledge, researchers can advance development and utilization of LLMs in a more informed and effective manner.",
}
| Large Language Models (LLMs) have not only exhibited exceptional performance across various tasks, but also demonstrated sparks of intelligence. Recent studies have focused on assessing their capabilities on human exams and revealed their impressive competence in different domains. However, cognitive research on the overall knowledge structure of LLMs is still lacking. In this paper, based on educational diagnostic assessment method, we conduct an evaluation using MoocRadar, a meticulously annotated human test dataset based on Bloom Taxonomy. We aim to reveal the knowledge structures of LLMs and gain insights of their cognitive capabilities. This research emphasizes the significance of investigating LLMs{'} knowledge and understanding the disparate cognitive patterns of LLMs. By shedding light on models{'} knowledge, researchers can advance development and utilization of LLMs in a more informed and effective manner. | [
"Zhang, Zheyuan",
"Yu, Jifan",
"Li, Juanzi",
"Hou, Lei"
] | Exploring the Cognitive Knowledge Structure of Large Language Models: An Educational Diagnostic Assessment Approach | findings-emnlp.111 | 2310.08172 | [
""
] | https://huggingface.co/papers/2310.08172 | 1 | 0 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.112.bib | https://aclanthology.org/2023.findings-emnlp.112/ | @inproceedings{torres-futrell-2023-simpler,
title = "Simpler neural networks prefer subregular languages",
author = "Torres, Charles and
Futrell, Richard",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.112",
doi = "10.18653/v1/2023.findings-emnlp.112",
pages = "1651--1661",
abstract = "We apply a continuous relaxation of $L_0$ regularization (Louizos et al., 2017), which induces sparsity, to study the inductive biases of LSTMs. In particular, we are interested in the patterns of formal languages which are readily learned and expressed by LSTMs. Across a wide range of tests we find sparse LSTMs prefer subregular languages over regular languages and the strength of this preference increases as we increase the pressure for sparsity. Furthermore LSTMs which are trained on subregular languages have fewer non-zero parameters. We conjecture that this subregular bias in LSTMs is related to the cognitive bias for subregular language observed in human phonology which are both downstream of a simplicity bias in a suitable description language.",
}
| We apply a continuous relaxation of $L_0$ regularization (Louizos et al., 2017), which induces sparsity, to study the inductive biases of LSTMs. In particular, we are interested in the patterns of formal languages which are readily learned and expressed by LSTMs. Across a wide range of tests we find sparse LSTMs prefer subregular languages over regular languages and the strength of this preference increases as we increase the pressure for sparsity. Furthermore LSTMs which are trained on subregular languages have fewer non-zero parameters. We conjecture that this subregular bias in LSTMs is related to the cognitive bias for subregular language observed in human phonology which are both downstream of a simplicity bias in a suitable description language. | [
"Torres, Charles",
"Futrell, Richard"
] | Simpler neural networks prefer subregular languages | findings-emnlp.112 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.113.bib | https://aclanthology.org/2023.findings-emnlp.113/ | @inproceedings{liu-etal-2023-simple,
title = "Simple Hardware-Efficient {PCFG}s with Independent Left and Right Productions",
author = "Liu, Wei and
Yang, Songlin and
Kim, Yoon and
Tu, Kewei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.113",
doi = "10.18653/v1/2023.findings-emnlp.113",
pages = "1662--1669",
abstract = "Scaling dense PCFGs to thousands of nonterminals via low-rank parameterizations of the rule probability tensor has been shown to be beneficial for unsupervised parsing. However, PCFGs scaled this way still perform poorly as a language model, and even underperform similarly-sized HMMs. This work introduces $\emph{SimplePCFG}$, a simple PCFG formalism with independent left and right productions. Despite imposing a stronger independence assumption than the low-rank approach, we find that this formalism scales more effectively both as a language model and as an unsupervised parser. We further introduce $\emph{FlashInside}$, a hardware IO-aware implementation of the inside algorithm for efficiently scaling simple PCFGs. Through extensive experiments on multiple grammar induction benchmarks, we validate the effectiveness of simple PCFGs over low-rank baselines.",
}
| Scaling dense PCFGs to thousands of nonterminals via low-rank parameterizations of the rule probability tensor has been shown to be beneficial for unsupervised parsing. However, PCFGs scaled this way still perform poorly as a language model, and even underperform similarly-sized HMMs. This work introduces $\emph{SimplePCFG}$, a simple PCFG formalism with independent left and right productions. Despite imposing a stronger independence assumption than the low-rank approach, we find that this formalism scales more effectively both as a language model and as an unsupervised parser. We further introduce $\emph{FlashInside}$, a hardware IO-aware implementation of the inside algorithm for efficiently scaling simple PCFGs. Through extensive experiments on multiple grammar induction benchmarks, we validate the effectiveness of simple PCFGs over low-rank baselines. | [
"Liu, Wei",
"Yang, Songlin",
"Kim, Yoon",
"Tu, Kewei"
] | Simple Hardware-Efficient PCFGs with Independent Left and Right Productions | findings-emnlp.113 | 2310.14997 | [
"https://github.com/sustcsonglin/TN-PCFG"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.114.bib | https://aclanthology.org/2023.findings-emnlp.114/ | @inproceedings{tian-etal-2023-r3,
title = "R$^3$ Prompting: Review, Rephrase and Resolve for Chain-of-Thought Reasoning in Large Language Models under Noisy Context",
author = "Tian, Qingyuan and
Zhu, Hanlun and
Wang, Lei and
Li, Yang and
Lan, Yunshi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.114",
doi = "10.18653/v1/2023.findings-emnlp.114",
pages = "1670--1685",
abstract = "With the help of Chain-of-Thought (CoT) prompting, Large Language Models (LLMs) have achieved remarkable performance on various reasoning tasks. However, most of them have been evaluated under noise-free context and the dilemma for LLMs to produce inaccurate results under the noisy context has not been fully investigated. Existing studies utilize trigger sentences to encourage LLMs to concentrate on the relevant information but the trigger has limited effect on final answer prediction. Inspired by interactive CoT method, where intermediate reasoning steps are promoted by multiple rounds of interaction between users and LLMs, we propose a novel prompting method, namely R$^3$ prompting, for CoT reasoning under noisy context. Specifically, R$^3$ prompting interacts with LLMs to perform key sentence extraction, variable declaration and answer prediction, which corresponds to a thought process of reviewing, rephrasing and resolving. The responses generated at the last interaction will perform as hints to guide toward the responses of the next interaction. Our experiments show that R$^3$ prompting significantly outperforms existing CoT prompting methods on five reasoning tasks under noisy context. With GPT-3.5-turbo, we observe 3.7{\%} accuracy improvement on average on the reasoning tasks under noisy context compared to the most competitive prompting baseline. More analyses and ablation studies show the robustness and generalization of R$^3$ prompting method in solving reasoning tasks in LLMs under noisy context.",
}
| With the help of Chain-of-Thought (CoT) prompting, Large Language Models (LLMs) have achieved remarkable performance on various reasoning tasks. However, most of them have been evaluated under noise-free context and the dilemma for LLMs to produce inaccurate results under the noisy context has not been fully investigated. Existing studies utilize trigger sentences to encourage LLMs to concentrate on the relevant information but the trigger has limited effect on final answer prediction. Inspired by interactive CoT method, where intermediate reasoning steps are promoted by multiple rounds of interaction between users and LLMs, we propose a novel prompting method, namely R$^3$ prompting, for CoT reasoning under noisy context. Specifically, R$^3$ prompting interacts with LLMs to perform key sentence extraction, variable declaration and answer prediction, which corresponds to a thought process of reviewing, rephrasing and resolving. The responses generated at the last interaction will perform as hints to guide toward the responses of the next interaction. Our experiments show that R$^3$ prompting significantly outperforms existing CoT prompting methods on five reasoning tasks under noisy context. With GPT-3.5-turbo, we observe 3.7{\%} accuracy improvement on average on the reasoning tasks under noisy context compared to the most competitive prompting baseline. More analyses and ablation studies show the robustness and generalization of R$^3$ prompting method in solving reasoning tasks in LLMs under noisy context. | [
"Tian, Qingyuan",
"Zhu, Hanlun",
"Wang, Lei",
"Li, Yang",
"Lan, Yunshi"
] | R^3 Prompting: Review, Rephrase and Resolve for Chain-of-Thought Reasoning in Large Language Models under Noisy Context | findings-emnlp.114 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.115.bib | https://aclanthology.org/2023.findings-emnlp.115/ | @inproceedings{deoghare-etal-2023-quality,
title = "Quality Estimation-Assisted Automatic Post-Editing",
author = "Deoghare, Sourabh and
Kanojia, Diptesh and
Blain, Fred and
Ranasinghe, Tharindu and
Bhattacharyya, Pushpak",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.115",
doi = "10.18653/v1/2023.findings-emnlp.115",
pages = "1686--1698",
abstract = "Automatic Post-Editing (APE) systems are prone to over-correction of the Machine Translation (MT) outputs. While Word-level Quality Estimation (QE) system can provide a way to curtail the over-correction, a significant performance gain has not been observed thus far by utilizing existing APE and QE combination strategies. In this paper, we propose joint training of a model on APE and QE tasks to improve the APE. Our proposed approach utilizes a multi-task learning (MTL) methodology, which shows significant improvement while treating both tasks as a {`}bargaining game{'} during training. Moreover, we investigate various existing combination strategies and show that our approach achieves state-of-the-art performance for a {`}distant{'} language pair, viz., English-Marathi. We observe an improvement of 1.09 TER and 1.37 BLEU points over a baseline QE-Unassisted APE system for English-Marathi, while also observing 0.46 TER and 0.62 BLEU points for English-German. Further, we discuss the results qualitatively and show how our approach helps reduce over-correction, thereby improving the APE performance. We also observe that the degree of integration between QE and APE directly correlates with the APE performance gain. We release our code and models publicly.",
}
| Automatic Post-Editing (APE) systems are prone to over-correction of the Machine Translation (MT) outputs. While Word-level Quality Estimation (QE) system can provide a way to curtail the over-correction, a significant performance gain has not been observed thus far by utilizing existing APE and QE combination strategies. In this paper, we propose joint training of a model on APE and QE tasks to improve the APE. Our proposed approach utilizes a multi-task learning (MTL) methodology, which shows significant improvement while treating both tasks as a {`}bargaining game{'} during training. Moreover, we investigate various existing combination strategies and show that our approach achieves state-of-the-art performance for a {`}distant{'} language pair, viz., English-Marathi. We observe an improvement of 1.09 TER and 1.37 BLEU points over a baseline QE-Unassisted APE system for English-Marathi, while also observing 0.46 TER and 0.62 BLEU points for English-German. Further, we discuss the results qualitatively and show how our approach helps reduce over-correction, thereby improving the APE performance. We also observe that the degree of integration between QE and APE directly correlates with the APE performance gain. We release our code and models publicly. | [
"Deoghare, Sourabh",
"Kanojia, Diptesh",
"Blain, Fred",
"Ranasinghe, Tharindu",
"Bhattacharyya, Pushpak"
] | Quality Estimation-Assisted Automatic Post-Editing | findings-emnlp.115 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.116.bib | https://aclanthology.org/2023.findings-emnlp.116/ | @inproceedings{bhardwaj-etal-2023-adapter,
title = "Adapter Pruning using Tropical Characterization",
author = "Bhardwaj, Rishabh and
Vaidya, Tushar and
Poria, Soujanya",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.116",
doi = "10.18653/v1/2023.findings-emnlp.116",
pages = "1699--1706",
abstract = "Adapters are widely popular parameter-efficient transfer learning approaches in natural language processing that insert trainable modules in between layers of a pre-trained language model. Apart from several heuristics, however, there has been a lack of studies analyzing the optimal number of adapter parameters needed for downstream applications. Thus, we propose an adapter pruning approach by studying the tropical characteristics of trainable modules. We cast it as an optimization problem that aims to prune parameters from the adapter layers without changing the orientation of underlying tropical hypersurfaces. Our experiments on five NLP datasets show that tropical geometry tends to identify more relevant parameters to prune when compared with the magnitude-based baseline, while a combined approach works best across the tasks.",
}
| Adapters are widely popular parameter-efficient transfer learning approaches in natural language processing that insert trainable modules in between layers of a pre-trained language model. Apart from several heuristics, however, there has been a lack of studies analyzing the optimal number of adapter parameters needed for downstream applications. Thus, we propose an adapter pruning approach by studying the tropical characteristics of trainable modules. We cast it as an optimization problem that aims to prune parameters from the adapter layers without changing the orientation of underlying tropical hypersurfaces. Our experiments on five NLP datasets show that tropical geometry tends to identify more relevant parameters to prune when compared with the magnitude-based baseline, while a combined approach works best across the tasks. | [
"Bhardwaj, Rishabh",
"Vaidya, Tushar",
"Poria, Soujanya"
] | Adapter Pruning using Tropical Characterization | findings-emnlp.116 | 2310.19232 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.117.bib | https://aclanthology.org/2023.findings-emnlp.117/ | @inproceedings{ikbal-etal-2023-self,
title = "Self-Supervised Rule Learning to Link Text Segments to Relational Elements of Structured Knowledge",
author = "Ikbal, Shajith and
Sharma, Udit and
Karanam, Hima and
Neelam, Sumit and
Luss, Ronny and
Sreedhar, Dheeraj and
Kapanipathi, Pavan and
Khan, Naweed and
Erwin, Kyle and
Makondo, Ndivhuwo and
Abdelaziz, Ibrahim and
Fokoue, Achille and
Gray, Alexander and
Crouse, Maxwell and
Chaudhury, Subhajit and
Subramanian, Chitra",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.117",
doi = "10.18653/v1/2023.findings-emnlp.117",
pages = "1707--1718",
abstract = "We present a neuro-symbolic approach to self-learn rules that serve as interpretable knowledge to perform relation linking in knowledge base question answering systems. These rules define natural language text predicates as a weighted mixture of knowledge base paths. The weights learned during training effectively serve the mapping needed to perform relation linking. We use popular masked training strategy to self-learn the rules. A key distinguishing aspect of our work is that the masked training operate over logical forms of the sentence instead of their natural language text form. This offers opportunity to extract extended context information from the structured knowledge source and use that to build robust and human readable rules. We evaluate accuracy and usefulness of such learned rules by utilizing them for prediction of missing kinship relation in CLUTRR dataset and relation linking in a KBQA system using SWQ-WD dataset. Results demonstrate the effectiveness of our approach - its generalizability, interpretability and ability to achieve an average performance gain of 17{\%} on CLUTRR dataset.",
}
| We present a neuro-symbolic approach to self-learn rules that serve as interpretable knowledge to perform relation linking in knowledge base question answering systems. These rules define natural language text predicates as a weighted mixture of knowledge base paths. The weights learned during training effectively serve the mapping needed to perform relation linking. We use popular masked training strategy to self-learn the rules. A key distinguishing aspect of our work is that the masked training operate over logical forms of the sentence instead of their natural language text form. This offers opportunity to extract extended context information from the structured knowledge source and use that to build robust and human readable rules. We evaluate accuracy and usefulness of such learned rules by utilizing them for prediction of missing kinship relation in CLUTRR dataset and relation linking in a KBQA system using SWQ-WD dataset. Results demonstrate the effectiveness of our approach - its generalizability, interpretability and ability to achieve an average performance gain of 17{\%} on CLUTRR dataset. | [
"Ikbal, Shajith",
"Sharma, Udit",
"Karanam, Hima",
"Neelam, Sumit",
"Luss, Ronny",
"Sreedhar, Dheeraj",
"Kapanipathi, Pavan",
"Khan, Naweed",
"Erwin, Kyle",
"Makondo, Ndivhuwo",
"Abdelaziz, Ibrahim",
"Fokoue, Achille",
"Gray, Alex",
"er",
"Crouse, Maxwell",
"Chaudhury, Subhajit",
"Subramanian, Chitra"
] | Self-Supervised Rule Learning to Link Text Segments to Relational Elements of Structured Knowledge | findings-emnlp.117 | null | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.118.bib | https://aclanthology.org/2023.findings-emnlp.118/ | @inproceedings{gehrmann-etal-2023-tata,
title = "{T}a{TA}: A Multilingual Table-to-Text Dataset for {A}frican Languages",
author = "Gehrmann, Sebastian and
Ruder, Sebastian and
Nikolaev, Vitaly and
Botha, Jan and
Chavinda, Michael and
Parikh, Ankur and
Rivera, Clara",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.118",
doi = "10.18653/v1/2023.findings-emnlp.118",
pages = "1719--1740",
abstract = "Existing data-to-text generation datasets are mostly limited to English. To address this lack of data, we create Table-to-Text in African languages (TaTA), the first large multilingual table-to-text dataset with a focus on African languages. We created TaTA by transcribing figures and accompanying text in bilingual reports by the Demographic and Health Surveys Program, followed by professional translation to make the dataset fully parallel. TaTA includes 8,700 examples in nine languages including four African languages (Hausa, Igbo, Swahili, and Yor{\`u}b{\'a}) and a zero-shot test language (Russian). We additionally release screenshots of the original figures for future research on multilingual multi-modal approaches. Through an in-depth human evaluation, we show that TaTA is challenging for current models and that less than half the outputs from an mT5-XXL-based model are understandable and attributable to the source data. Our results highlight a) the need for validating metrics; and b) the importance of domain-specific metrics.",
}
| Existing data-to-text generation datasets are mostly limited to English. To address this lack of data, we create Table-to-Text in African languages (TaTA), the first large multilingual table-to-text dataset with a focus on African languages. We created TaTA by transcribing figures and accompanying text in bilingual reports by the Demographic and Health Surveys Program, followed by professional translation to make the dataset fully parallel. TaTA includes 8,700 examples in nine languages including four African languages (Hausa, Igbo, Swahili, and Yor{\`u}b{\'a}) and a zero-shot test language (Russian). We additionally release screenshots of the original figures for future research on multilingual multi-modal approaches. Through an in-depth human evaluation, we show that TaTA is challenging for current models and that less than half the outputs from an mT5-XXL-based model are understandable and attributable to the source data. Our results highlight a) the need for validating metrics; and b) the importance of domain-specific metrics. | [
"Gehrmann, Sebastian",
"Ruder, Sebastian",
"Nikolaev, Vitaly",
"Botha, Jan",
"Chavinda, Michael",
"Parikh, Ankur",
"Rivera, Clara"
] | TaTA: A Multilingual Table-to-Text Dataset for African Languages | findings-emnlp.118 | null | [
"https://github.com/google-research/url-nlp"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.119.bib | https://aclanthology.org/2023.findings-emnlp.119/ | @inproceedings{tang-etal-2023-explain,
title = "Explain-then-translate: an analysis on improving program translation with self-generated explanations",
author = "Tang, Zilu and
Agarwal, Mayank and
Shypula, Alexander and
Wang, Bailin and
Wijaya, Derry and
Chen, Jie and
Kim, Yoon",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.119",
doi = "10.18653/v1/2023.findings-emnlp.119",
pages = "1741--1788",
abstract = "This work explores the use of self-generated natural language explanations as an intermediate step for code-to-code translation with language models. Across three types of explanations and 19 programming languages constructed from the MultiPL-E dataset, we find the explanations to be particularly effective in the zero-shot case, improving performance by 12{\%} on average. Improvements with natural language explanations are particularly pronounced on difficult programs. We release our dataset, code, and canonical solutions in all 19 languages.",
}
| This work explores the use of self-generated natural language explanations as an intermediate step for code-to-code translation with language models. Across three types of explanations and 19 programming languages constructed from the MultiPL-E dataset, we find the explanations to be particularly effective in the zero-shot case, improving performance by 12{\%} on average. Improvements with natural language explanations are particularly pronounced on difficult programs. We release our dataset, code, and canonical solutions in all 19 languages. | [
"Tang, Zilu",
"Agarwal, Mayank",
"Shypula, Alex",
"er",
"Wang, Bailin",
"Wijaya, Derry",
"Chen, Jie",
"Kim, Yoon"
] | Explain-then-translate: an analysis on improving program translation with self-generated explanations | findings-emnlp.119 | 2311.07070 | [
"https://github.com/pootiet/explain-then-translate"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.